Servo 2008 - 02

V
o
l
.

6

N
o
.

2
S
E
R
V
O
M
A
G
A
Z
I
N
E
R
O
B
O
T
S

A
T

M
A
K
E
R

F
A
I
R
E
P
E
R
S
O
N
A
L

R
O
B
O
T
S
M
E
C
H
B
A
S
H
B
A
T
T
E
R
Y

C
A
P
A
C
I
T
Y
F
e
b
r
u
a
r
y

2
0
0
8
Cover.qxd 1/10/2008 1:04 PM Page 1
Full Page.qxd 1/7/2008 1:46 PM Page 2
4 SERVO 02.2008
SERVO Magazine (ISSN 1546-0592/CDN Pub Agree#40702530)
is published monthly for $24.95 per year by T & L Publications, Inc., 430
Princeland Court, Corona, CA 92879. PERIODICALS POSTAGE PAID
AT CORONA, CA AND AT ADDITIONAL ENTRY MAILING OFFICES.
POSTMASTER: Send address changes to SERVO Magazine, P.O.
Box 15277, North Hollywood, CA 91615 or Station A, P.O.
Box 54, Windsor ON N9A 6J5; cpcreturns@servomagazine.com
Departments
06 Mind/Iron
20 New Products
30 Events Calendar
53 Robotics Showcase
56 Robo-Links
73 SERVO Webstore
82 Advertisers Index
Columns
08
Robytes by Jeff Eckert
Stimulating Robot Tidbits
10
GeerHead by David Geer
Robot Leaves Breadcrumbs
14
Ask Mr. Roboto by Pete Miles
Your Problems Solved Here
62
Robotics Resources
by Gordon McComb
The Recycled Robot
67
Lessons From The Lab
by James Isom
NXT Packbot: Part 3
76
Appetizer by Kevin Berry
The Door Into Spring
78
Then and Now by Tom Carroll
Personal Robots: From Science Fiction
To Reality
PAGE 48
PAGE 10
TOC Feb08.qxd 1/10/2008 10:51 AM Page 4
02.2008
VOL. 6 NO. 2
SERVO 02.2008 5
ENTER WITH CAUTION!
22 The Combat Zone
31 The RoboCooler
by Jerry Reed
Find a few design tidbits in this
embedded appliance application
to use in your next build.
39 The MechBash Traveling
Robot Show
by Jon Vandervelde
See what happens when you combine
Mechwars robot combat with BotBash
into one event.
43 Maker Faire
by R. Steven Rainwater
Pleos, horses, monkeys, and parrots
were just a few of the robotic pets
that Faire attendees had the
opportunity to get up close and
personal with.
48 Building a Stepper Motor
Controller: Part 1
by Fred Eady
This controller is based on the
STMicroelectronics L6208, which is a
single-IC DMOS driver.
54 Capacity: The Key to
Battery Runtime
by Isidor Buchmann
Take a quick look at emerging
rapid-test technologies for
deep-cycle, lead-acid batteries.
57 Reviving an Androbot BOB
by Robert Doerr
Get your BOB rolling again.
Features & Projects
PAGE 43
TOC Feb08.qxd 1/10/2008 10:53 AM Page 5
Published Monthly By
T & L Publications, Inc.
430 Princeland Court
Corona, CA 92879-1300
(951) 371-8497
FAX (951) 371-3052
Webstore Only 1-800-783-4624
www.servomagazine.com
Subscriptions
Toll Free 1-877-525-2539
Outside US 1-818-487-4545
P.O. Box 15277
North Hollywood, CA 91615
PUBLISHER
Larry Lemieux
publisher@servomagazine.com
ASSOCIATE PUBLISHER/
VP OF SALES/MARKETING
Robin Lemieux
display@servomagazine.com
EDITOR
Bryan Bergeron
techedit-servo@yahoo.com
CONTRIBUTING EDITORS
Jeff Eckert Tom Carroll
Gordon McComb David Geer
Pete Miles R. Steven Rainwater
Fred Eady Kevin Berry
James Isom Robert Doerr
Isidor Buchmann Jon Vandervelde
Jerry Reed Russ Barrow
Chris Olin
CIRCULATION DIRECTOR
Tracy Kerley
subscribe@servomagazine.com
MARKETING COORDINATOR
WEBSTORE
Brian Kirkpatrick
sales@servomagazine.com
WEB CONTENT
Michael Kaudze
website@servomagazine.com
PRODUCTION/GRAPHICS
Shannon Lemieux
Joe Keungmanivong
ADMINISTRATIVE ASSISTANT
Debbie Stauffacher
Copyright 2008 by
T & L Publications, Inc.
All Rights Reserved
All advertising is subject to publishers approval.
We are not responsible for mistakes, misprints,
or typographical errors. SERVO Magazine
assumes no responsibility for the availability or
condition of advertised items or for the honesty
of the advertiser. The publisher makes no claims
for the legality of any item advertised in SERVO.
This is the sole responsibility of the advertiser.
Advertisers and their agencies agree to
indemnify and protect the publisher from any
and all claims, action, or expense arising from
advertising placed in SERVO. Please send all
editorial correspondence, UPS, overnight mail,
and artwork to: 430 Princeland Court,
Corona, CA 92879.
REFLEX CONTROL
Although fully autonomous
robots are often viewed as the
pinnacle of robotics, sometimes its
desirable for robots to follow specific
directions, under direct user control.
For example, a surgeon controlling a
surgical robot may want to exactly
specify the location of an incision.
The cost of full user control is the
demand placed on the operator. As
experienced with the MQ-1 Predator
and other drone combat aircraft has
shown, the control of a single robot
can require the full attention of
several humans the basic crew for a
Predator is one pilot and two sensor
operators. Similarly, if youve ever
controlled a mobile or air-borne robot
with an R/C unit through a video link
or even an R/C battle-bot in direct
sight you know that the task
requires focus and concentration,
leaving little room for other activities.
Controlling robots wearable or
otherwise through neuromuscular
signals is an obvious, albeit challenging,
approach to freeing the operator to
focus on other tasks. Instead of handling
a joystick, the operator simply moves an
arm or leg and the exoskeleton follows,
using the electrical signals activating the
muscles as a trigger. (Watch for Kazuo
Kiguchis article on Control of ExoSuits
with Biological Signals coming in the
March issue.)
One of the issues in biological
control of robots is how to handle
reflexes semi-autonomous movements
that are not consciously directed. To
understand the relevance of reflexes to
robotics, consider your normal reflexes.
If you accidentally brush your hand
against the hot tip of a soldering iron,
your hand will instinctively and
without conscious control instantly
recoil from the heat source. If you are
fast enough, you might get away with a
minor reddening of the skin that quickly
dissipates. Without the reflex, youd be
left with a serious burn, possible
infection, and a permanent scar.
The reflex arc illustrated in Figure 1
involves sensors in your skin and muscle,
a neurological pathway from the sensors
to your spinal cord, a connection within
the cord to the outbound motor neurons
controlling muscles in your hand and
arm, and neuromuscular innervation.
The brain isnt normally part of a reflex.
When it is involved, its usually to inhibit
the reflex you dont want to rip your
arm away from a nurse giving you a
vaccination, for example.
A practical advantage of a reflex
action is speed. Conscious control
involving hundreds or thousands of
neural synapses is simply slower than a
reflex arc involving a few neurons. If
you had to consciously blink every time
something headed for you eyes, youd
probably be blind by now. Another
advantage of reflexes is that they
enable you to avoid danger or at least
minimize damage while maintaining
conscious focus on the task at hand.
Given the advantage of near
instantaneous reflexes in biological
systems, its reasonable to assume that
exoskeletons and other forms of robots
can benefit from similar capabilities.
However, a dilemma faced by developers
of wearable robotic systems is the
degree to which the wearer should be
insulated from the environment.
At one extreme, the wearer is
unaware of extreme heat, impact, and
Mind / Iron
by Bryan Bergeron, Editor
Mind/Iron Continued
6 SERVO 02.2008
Mind-Feed Feb08.qxd 1/10/2008 11:09 AM Page 6
other dangers in the environment. Its up to exoskeletons
and built-in protective reflexes to keep the wearer from
danger. The downside to this scenario is that the wearer
would have to adapt to sudden reflex movements of the
exoskeleton, and the movements would have to be
controlled to avoid joint or muscle damage. At the other
extreme, the wearer of a robotic exoskeleton would be
exposed to whatever forces contact the exoskeleton, with
fidelity determined by the available sensors and transducers.
Somewhere in the middle seems reasonable for many
applications. For example, a soldier with a bulletproof
robotic exoskeleton should be protected from the physical
damage of bullets or shrapnel, but not necessarily from a
modest sting of impact. The modest pain would result in
biological reflexes, and the exoskeleton could respond to
the associated neuromuscular activity. Pain is often a
reliable indicator that something is wrong and that action
is necessary to avoid injury.
The impact of shrapnel or a bullet on a soldiers
bulletproof exoskeleton could be translated into a mildly
uncomfortable sensation on the wearers skin just beneath
the point of impact. Along this line of reasoning, variants of
Immersions TouchSense technology (www.immersion.
com/industrial/touchscreen/) could be used to synthesize
sensations through tactile transducers excited by signals of
varying frequency, intensity, duration, and wave shapes.
Tactile synthesis for wearers of sensor-studded
exoskeletons, socks, or shoes has potential medical value, as
well. Wearable sensors coupled with tactile synthesizers can
replace and supplant the desensitized peripheral sensory
organs of diabetics and other sufferers of peripheral
neuropathy. Its not unusual for sufferers of diabetes to tear
a toenail or cut their foot, only to discover the damage hours
later, when they happen to notice a blood-soaked sock. A
thin, lightweight, flexible sensory exoskeleton that generates
synthetic tactile feedback on the wearers back or other area
not affected by neuropathy could provide life-saving feedback
to diabetics, as well as advance the field of robotics.
If you want to take up the challenge, start by exploring
the computer literature on virtual reality and the psychological
literature on perception and the synthesis of sensation. SV
SERVO 02.2008 7
Spinal Cord
Sensory Motor
Muscle
Heat
FIGURE 1.
Reflex Arc.
P
erform proportional speed, direction, and steering with
only two Radio/Control channels for vehicles using two
separate brush-type electric motors mounted right and left
with our mixing RDFR dual speed control. Used in many
successful competitive robots. Single joystick operation: up
goes straight ahead, down is reverse. Pure right or left twirls
vehicle as motors turn opposite directions. In between stick
positions completely proportional. Plugs in like a servo to
your Futaba, JR, Hitec, or similar radio. Compatible with gyro
steering stabilization. Various volt and amp sizes available.
The RDFR47E 55V 75A per motor unit pictured above.
www.vantec.com
STEER WINNING ROBOTS
WITHOUT SERVOS!
Order at
(888) 929-5055
Mind-Feed Feb08.qxd 1/10/2008 11:06 AM Page 7
8 SERVO 02.2008
UAV Imitates Sea Birds
So one day Guy Meadows, director
of the Marine Hydrodynamics Labs at
the University of Michigan (www.
umich.edu), was floating around and
saw a flying fish pop out of the water
and soar over the waves. He was so
impressed and inspired that he said,
Wow. Ill bet I can build one of those,
hence the name of the UAV that he and
some engineering researchers designed
and built. Somehow the concept
evolved away from fish and focused on
sea birds, but the name stuck.
In any event, Meadows and his
colleagues did a study of things that
go flap and discovered that many of
them have some traits in common,
such as weighing about 20 lb and
having a 2 m wingspan. It turns out
that this is pretty much the ideal
aerodynamic design for skimming
close to the surface, so the Flying Fish
is physically similar to a large, mechan-
ical pelican. It is also believed to be the
first seaplane that can initiate and
perform its own takeoffs and landings.
It may sound like all fun and
games, and a good excuse to escape
Ann Arbor for some quality time in
Monterey, CA, testing the thing, but
project funding came from the DoDs
Defense Advanced Research Projects
Agency (DARPA), with the aim of
advancing the agencys Persistent
Ocean Surveillance program.
In operation, the electric-powered
UAV drifts along until its onboard GPS
tells it that it has floated too far. The
takeoff sequence is then triggered, and
Flying Fish goes airborne in about 10 m.
When it reaches the proper GPS coordi-
nates, it lands using a shallow descent.
The next step will be to fit the plane with
solar power and an array of sensors.
Bots to Fight Fires in
London
Apparently, there are a serious
number of incidents in the London
area involving fire, acetylene cylinders,
and railroad tracks. Such incidents
cause great consternation among rail
travelers, as the lines have to be shut
down until the danger has been
eliminated. As a result, Network Rail, in
conjunction with the London Fire
Brigade, has commissioned QinetiQ Ltd.
to provide and operate some specialized
firefighting ROVs for a six-month trial.
The plan is to use the bots cameras
to identify whether any acetylene
cylinders are present when fire
breaks out near the tracks and
using thermal imaging gauge
whether such cylinders have
cooled off enough to allow human
firefighters to approach them.
Three types of ROVs are
included in the trial: Talon, a
small tracked vehicle used in Iraq
for bomb disposal and here fitted
with video and thermal image
cameras; Black Max, basically a
squirtbot; and Brokk 90, a heavy-
duty mini-digger designed to
remove debris and gain access to
burning vehicles and structures.
According to a company
spokesman, QinetiQ has already been
called to deploy the ROVs on a number
of occasions, and they have each
proved useful in assisting the Fire
Brigade in dealing with the incidents.
Sounds like a bit of all right.
BigDog on the Block
Also sponsored by the folks at
DARPA is BigDog, billed as the
most advanced quadruped robot on
Earth. Built by Boston Dynamics, it
is a major part of the agencys
Biodynotics (biologically inspired
dynamic robots) program, which aims
to apply biological principles to devel-
op robots that can better move
through difficult terrain, travel more
efficiently, and recover from stumbles.
The BigDog portion of the
program aims to replace tracked and
wheeled systems with legged ones,
eventually demonstrating mule-sized,
200-lb platforms that can carry
payloads of supplies, ammunition,
weapons, and other items that soldiers
now have to tote.
At present, BigDog measures 1 m
(3.3 feet) long, is 0.7 m (2.3 feet) tall,
and weighs 75 kg (165 lb). The dog
food this guy consumes is gasoline,
which feeds an engine that drives
the hydraulic actuation system. An on-
The Flying Fish UAV. Photo courtesy
of the University of Michigan.
ROVs fire robots in action.
Photo courtesy of QinetiQ Ltd.
The BigDog quadruped robot. Photo courtesy
of the US Department of Defense.
by Jeff Eckert
Robytes.qxd 1/10/2008 10:34 AM Page 8
board computer controls locomotion,
the legs, and a complex variety of
sensors. So far, BigDog has trotted 5.3
km/hr (3.3 mph), climbed a 35 slope,
and carried a 54-kg (120-lb) load. For a
fascinating video, visit www.boston
dynamics.com/content/sec.php?
section=BigDog.
Raven Takes Flight
It might look like a model airplane
that you would fly in the park on
Sunday, but the RQ-11B Raven is a
serious little cousin of the MQ-1
Predator and MQ-9 Reaper. The little
UAV weighs only about 4 lb and has
just a 55-in (1.4-m) wingspan, but it
can conduct visual reconnaissance up
to 10 miles from its launch point and
climb to 10,000 feet above sea level.
It sends live footage back to the
operator for later evaluation.
According to a spokesman,
the Raven B is particularly
useful during convoy operations,
because it can travel at 17 to 44
knots and keep up with most
trucks. It is also good for target
acquisitioning, battle damage
assessment, and detection assess-
ment for ground-based threats.
Raven carries a camera that
provides high-res imagery up to
500 feet above the terrain, with
location coordinates shown on
the display. It can remain air-
borne for 1.5 hours on a charge.
Perhaps its most interesting
characteristic is that it has no landing
gear; it is designed to break apart,
undamaged, on impact, and it can be
reassembled in a matter of minutes.
The exact cost of the little bird was
not disclosed, but the Danish Army
recently ordered 12 systems (including
logistics support and training) at a cost
of $2.4 million, which would price it at
$200,000. Yes, nothing youll be flying
in the park for fun. SV
Robyt es
Staff Sgt. Marie Garcia launches an RQ-11B
Raven at Bagram Air Base, Afghanistan. US
Air Force photo by Staff Sgt. Mike Andriacco.
SERVO 02.2008 9
Robytes.qxd 1/10/2008 1:13 PM Page 9
10 SERVO 02.2008
M
oravian College student and
roboticist Wesley Moser (class
of 08) built a robot that
could trace its steps and map them out
on a computer screen, albeit with a lot
of help from Mosers own software,
which he programmed using multiple
languages.
The robot was the result of
Mosers Student Opportunity for
Academic Research (SOAR) project at
Moravian. Ben Coleman, assistant
professor of computer science at the
academic institution, guided Moser.
The robot uses a variety of sensors
to traverse the boundaries of its given
landscape; in this case, a box. Along
with the ability to avoid obstacles
without which it would be quite clumsy
its sensors and computer technology
enable it to record where it has been.
The robot then distributes the
information to a computer, which draws
a line duplicating the robots path.
Mapping and
Maneuvering the
Robots Intelligence
Technically, the robot itself doesnt
actually know where it has been,
according to Moser. As the robot rolls
around inside its little playpen, it
keeps a heading. Every few parts of a
second, it sends its heading back to a
computer (laptop).
The heading is simply information
about whether it is turning or not and
in which direction, according to Moser.
Along with the heading, the robot
returns its recent sensor readings and
gives the computer the exact time
since its last update, Moser explains.
The computer then compiles all of
these readings in real-time and
generates a single-line mapping
of where the robot has been. In
this case, the robot is really the
explorer and the laptop has to
do all the work of knowing where
the robot was, continues Moser.
The robot houses its
intelligence in a Handy Board. The
Handy Board consists of a
Motorola MC68HC11 processor,
32K of RAM, and a variety of sen-
sor inputs. It also supports four
motors. Despite its comparably
slow clock speed, the 2 MHz
processor was just speedy enough
to shoulder all the responsibilities
Moser had developed for it. We were
coming up on its limit when trying to
send data to the computer, collect data
readings, and determine the next
heading all at once, Moser notes.
Moser used a programming
language called Interactive-C to enable
the robots cranium so it could think.
Through this programming, the robot
could complete tasks like polling
sensors for data, controlling its motors,
and doing direct access calls to the
on-board RAM.
Moser first programmed the robot
with a path-finding ability, with both the
robot and sonar facing straight ahead.
But he quickly discovered that the robot
collided with walls when approaching
them from an angle. Moser needed the
robot to be able to see obstacles from all
sides, and so he added a servo to turn
the sonar side to side. Still, the robot
crashed into walls, so Moser added
infrared sensors so the robot could sense
when it was too close to the walls of its
box, whether the sonar agreed or not!
Once Moser had decided to save
the robots path and record it, he real-
ized he had to do more programming.
I added some programming routines
to get the robot to remember what it
had done, but the robot only had
enough memory to remember the
previous 30 seconds, Moser says.
Moser had to clear that memory
and move that data off the robot to
another computer to make room for
each new wave of information. While
Moser considered a wireless link to
Contact the author at geercom@alltel.net by David Geer
Robot Leaves Breadcrumbs
A trailblazer that traces its own steps
This is a front and side angle view of
the fully assembled robot.
Geerhead.qxd 1/10/2008 10:25 AM Page 10
GEERHEAD
move the data, this would be cost-
prohibitive and too much extra
processing work for the robots brain.
So, Moser wrote a program that
moved the data to the laptop via cable.
According to Moser, its easy to
program the Handy Board. He simply
wrote his code on a computer and
downloaded it to the board. Our partic-
ular Handy Board had an expansion
board on it, which allowed us to use the
sonar, the servo, and gave a few extra
sensor inputs. The Handy Board also has
a small, 16 x 2 LCD screen for displaying
the status of the robot, Moser says.
The board supports three infrared
sensors, a sonar and control box, and a
servomotor for turning the sonar.
The laptop runs a program of
Mosers design, written in C++ and
OpenGL. This program uses the
robots data to draw the map on the
computer screen.
The program parses the robots
data and determines where the robot
would be if it were situated on a
Cartesian coordinate plane. The
program assumes that the robot starts
positioned at (0,0). When the robot
sends its data, the program takes the
current direction and projects a circle in
front of the robot with a radius that is
dependent on how much the robot is
turning, says Moser.
Then, the program calculates the
next point on the plane or map (the
robots next location) by using the speed
of the robot and the time that has
passed to determine how far the robot
has traveled. The process repeats itself
starting from the newly plotted point.
The mapping process works better
in theory than in practice, due to a few
factors. First, the motors dont always
run at the same speed. So, using speed
and time to calculate the robots new
location is inaccurate. Speed also
varied at times as the batteries ran low.
An improvement to our design
would be to add an odometer to each
rear wheel to actually tell how far the
wheel has turned, but with our current
design the robot could not handle yet
another task it would slow down the
robots processing time too much,
Moser says.
Communications between the
robot and computer were particularly
challenging. The robot sent data across
a cable to the computer in the form of
a stream of characters. The computer
received and parsed the data in order
to interpret it.
But the computer had to deter-
mine when and whether it had
received a complete and uncorrupted
data stream before it could act on the
data. Occasionally the computer
would end up in an endless loop
collecting data because one part of the
data got corrupted, and the computer
was waiting for the rest of it. I suppose
this is par for the course with data
communication, but this was my first
experience with it, Moser explains.
Around, Not Through
The robot gets around objects
using its sonar. Another technician
created the sonar for Moser from a
Polaroid camera. I believe Polaroid used
this particular sonar to determine the
distance from their camera to the object
it was shooting, Moser suggests.
The sonar estimates the distance
between the robot and other objects. It
does a good job of estimating these dis-
tances and detecting objects if they are
between one and eight feet away from
the robot. The sonar turns 62 degrees
to the right or left of the robot to detect
objects that come within range on
either side, according to Moser.
However, the sonar is notoriously unre-
liable, as it will occasionally see some-
thing that is not there, Moser says.
Moser wrote some programming
routines to help the sonar see within
one foot of the robot, but at about six
inches away, the sonars capabilities
really fell off.
The way the sonar works, Moser
tells it to ping by setting a bit on the
Handy Board, and the sonar returns a
number between 0 and 32768 repre-
senting the distance away, with 0 being
SERVO 02.2008 11
This is the robot as it is about to be
disassembled. Here, the sonar speaker
has been removed from the servomotor.
Moser has his hand on the Handy Board
on top, about to peel it away from
the robot.
This mostly head-on view of the robot
reveals its sonar speaker and grill-like
covering up top (black). The robot uses this
speaker to send and receive sonar pings.
The nearly transparent box on the side
houses the sonar control box. This is the
mechanism that actually sends and receives
the pings through the speaker element.
Beneath the sonar speaker is a black servo
motor and housing, which looks a lot like
printer cartridges in form factor. But rather
than printing with ink, this motor turns the
sonar to map out obstacles in the robots
path. Directly beneath the servo motor
inside the blue rectangle of LEGOs, is the
front infrared sensor.
12 SERVO 02.2008
right in front (one foot away) and 32768
being infinity, or about eight feet away
for the robot/sonar, according to Moser.
To allow the sonar to see things clos-
er than one foot, Moser had to suppress
something called a blanking feature on
the sonar. The blanking feature prevents
the robot from hearing the exact ping
that it had just sent out before it hears
the echo that returns; in this way, it does-
nt recognize its own sound as being the
returning sound, which would negate
the distance calculations. By inhibiting
this blanking feature and implementing
my own (smaller) blanking interval, the
robot was able to detect things that
were closer to it, says Moser.
The robot also facilitates obstacle
detection and avoidance by use of its
three infrared sensors, which are locat-
ed on the front and to the right and
left. These infrared sensors can pick up
objects up to three inches away. They
also return a number within a range of
numbers to determine the distance the
object is away (in this case, the number
returned is between 0 and 255).
While these sensors take up the
slack for the sonar at the closest ranges,
they are less effective in sun light,
detecting the natural lights infrared rays
as being something close to the bot. I
wrote an algorithm to work around this,
which decreases the sensitivity of the
infrared sensors (or disables them com-
pletely) if it appears that the robot is sur-
rounded by natural light, says Moser.
The laptop computer combines the
input from the infrared sensors and the
sonar and weighs them when determin-
ing what path the robot should take. SV
GEERHEAD
Moravian College
www.moravian.edu
Handy Board microcontroller
www.handyboard.com
OpenGL programming language
www.opengl.org
A Handy Board and Interactive C
programming language resource
http://blackcat.brynmawr.edu/~
dkumar/UGAI/hb-ic.html
An Interactive C programming
language manual/resource
www.owlnet.rice.edu/~elec201/
Book/icmanual.html
A C++ programming language tutorial
www.cplusplus.com/doc/tutorial
RESOURCES
The path of the robot is projected
back into the image for us to see. As
the path of light demonstrates, the
robot frequently miscalculates the
distance between itself and the sides
of the box. As a result, it repeatedly
gets very close to things before turning
off in another direction.
Here is the robot nearly in pieces. At the
bottom is the interface box and cable.
Directly above it are the Handy Board
and expansion board. The expansion
board drives the sonar and offers a few
more sensor inputs for the robots use.
The sonar control box is visible with the
sonar speaker attached. I am holding the
frame of the robot, which is essentially
made of LEGOs. In the bottom of the
frame are two LEGO-style motors, which
drive the wheels through a series of gears,
says Moser.
14 SERVO 02.2008
Q
. This might sound like a stupid question, but how
do you display numbers on a serial LCD? I have a
serial LCD from Parallax that was given to me at
this years FLL (FIRST LEGO League) tournament, and it
works great for printing characters, but when I send it a
number it prints what looks like a random number. Yeah, I
know, its an ASCII character. So, how do you convert a
byte size number into a three digit number to display on the
serial LCD? I dont suppose that you would be willing to
explain how to convert that same byte into an eight digit
binary number?
Pat Smith
A
. Like I always say, there is no such thing as a stupid
question. The only reason you are asking is because
you dont know, and I am sure that there are a lot of
other people that have the same question in the back of their
minds. It is great to hear that you are participating in the
FIRST LEGO League (www.firstlegoleague.org). I think
everyone reading this magazine should get involved with the
different FIRST activities. The Parallax serial LCD (www.
parallax.com) is a really simple LCD module to get up and
running. Only one wire, plus power and ground is needed
to control it.
This isnt that difficult, and many microcontrollers
will automatically do this for you, such as the BASIC
Stamps from Parallax. Basically, all you have to do is take
your number and divide it by 100. The integer part of
the result is the first digit to be displayed. Then take the
remainder of that division and divide it by 10, and the
integer part of that result is the second digit to be
displayed. And what is remaining, is the third digit to be
displayed. The LCD_DEC programming example shown
here is a short and simple BASIC Stamp 2 program that
will display a single byte value as a three digit number on
the serial LCD.
TxPin CON 0
LCD_Baud CON 32 9600 baud on a BS2
temp1 VAR Byte
temp2 VAR Byte
temp3 VAR Byte
temp4 VAR Byte
i VAR Byte
temp1 = 195
LCD_DEC:
temp2=temp1/100
temp2 = temp2 + 48
Serout TxPin, LCD_Baud, [temp2]
temp2=temp1//100
temp3=temp2/10
temp3=temp3 + 48
temp4=temp2//10
temp4=temp4 + 48
Temp1 is some number that I have randomly chosen
to be displayed on the LCD. If you are not familiar with
the // function, it is called the MODULUS function and it
Tap into the sum of all human knowledge and get your questions answered here!
From software algorithms to material selection, Mr. Roboto strives to meet you
where you are and what more would you expect from a complex service droid?
by
Pete Miles
Our resident expert on all things
robotic is merely an Email away.
roboto@servomagazine.com
Figure 1. BlueSMiRF serial modem from SparkFun Electronics.
MrRoboto.qxd 1/10/2008 9:51 AM Page 14
returns the remainder of a division operation. For example,
in regular division, 195/100 = 1 (BASIC Stamps use
integer math, so all the decimal places are truncated/lost).
Now, 195//100 = 95. Then, 95/10 = 9 and 95//10 = 5.
Thus, these are the three digits that will be displayed 1,
9, and 5. The Parallax serial LCD doesnt actually display
numbers but rather ASCII characters. So to convert these
numbers into the appropriate ASCII character digits, the
decimal 48 is added to each digit. Forty eight is the
ASCII character for the number 0 (zero), and by adding 48
to the digit, it will map this digit to the appropriate
number character on the ASCII table which will then be
displayed properly.
Now, converting a single byte into an eight character
binary number requires reading all eight bits of the byte,
one at a time. To do this, a simple loop is used. Since
writing to the LCD is like writing by hand (from left
to right) each bit must be read from left to right (most
significant bit, MSB to the least significant bit, LSB) and
then displayed on the LCD in that order. The sample
routine, LCD_BIN, demonstrates how this is done. The key
function in this routine is the rotate left operator, <<. This
function will shift every bit in the byte to the left. Temp2
<< 1 means that every bit in the byte is shifted to the left
by one place. If temp2 << 2 was used, it would shift all the
bits by two places. The bits dont wrap around, thus the
temp2 variable is destroyed in the process of reading each
bit. This is why temp2 = temp1 is used. Temp1 is the
primary variable, and temp2 is just a copy of it so that the
original variable isnt changed in the process of reading
each of its bits. The IF-THEN statement checks the MSB of
temp2 and will display its result on the serial LCD. Thats all
there is to it.
LCD_BIN:
temp2 = temp1
FOR i = 1 TO 8
IF temp2.BIT7 = 1 THEN
SEROUT TxPin, LCD_Baud, [1]
ELSE
SEROUT TxPin, LCD_Baud, [0]
ENDIF
temp2 = temp2 << 1
NEXT
Now, just in case you were wondering how to display
hexadecimal numbers, the LCD_HEX routine shown here
demonstrates how to do this. This routine is very similar to
the LCD_DEC routine, except that instead of dividing the
number by 100 and 10, you are only dividing the number
once by 16. As with the LCD_DEC routine, the numbers
need to be mapped to the ASCII character set. Adding 48
to the decimal numbers will display the decimal characters.
When the number is greater than nine, i.e., 10, 11, 12,
etc., it is actually A, B, C, etc., in hexadecimal. To map
this properly, we add 55 to the numbers so that the prop-
er characters are displayed on the LCD. The conditional
IF-THEN statement in this routine performs the mapping
operation.
LCD_HEX:
temp2=temp1/16
IF temp2 > 9 THEN
temp2 = temp2 + 55
ELSE
temp2 = temp2 + 48
ENDIF
SEROUT TxPin, LCD_Baud, [temp2]
temp2=temp1//16
IF temp2 > 9 THEN
temp2 = temp2 + 55
ELSE
temp2 = temp2 + 48
ENDIF
SEROUT TxPin, LCD_Baud, [temp2]
The BASIC Stamp and many other microcontrollers have
special built-in functions that do this for you, but if your
microcontroller doesnt have them, then these routines will
show you how to do it. So, how would you write a routine to
display Octal numbers?
++++++++++++++++++++++++++++++++++++++++++++++
Q
. I am writing to you in hopes that you can show me
how to make my RoboNova robot a remote
controlled robot with something like a Playstation 2
controller. When I saw your article last October about
turning a Playstation 2 controller into a remote control, I got
excited about this. The Robonova robot has a lot of really
cool moves that cant be controlled with a regular aircraft
R/C system, but the Playstation 2 controller has 16 buttons
and two joysticks that can be used to control all the robots
functions. I dont like the IR remote that comes with it
because you practically have to be on top of the robot in
order for it to work, and then it only works sometimes. Any
help here would be greatly appreciated.
Lance Alberts
A
. Youre in luck! I recently did the same thing with my
robot. The work to prepare for the October 07 issue of
SERVO Magazine inspired me to see if I could convert
my RoboNova (www.hitecrobotics.com) into a remote
control robot. My goal was to see if I could get this robot to
compete in the remote control division of the 3 kg robot
sumo contest at Robothon (www.robothon.org) that is
hosted by the Seattle Robotics Society (www.seattlerobot
ics.org). Needless to say, it didnt win, but it sure was a
crowd pleaser. Especially when one of his opponents got
stuck between his legs a true wrestling match.
The BlueSMiRF serial modem (see Figure 1) from
SparkFun Electronics (www.sparkfun.com) makes this a
fairly straightforward upgrade. You will need two of them:
one for the robot and the other for your Playstation 2
controller. Figure 2 shows a schematic drawing for wiring the
BlueSMiRF serial modem to a SX28 microcontroller from
Parallax. Figure 3 shows a schematic cable for connecting
the BlueSMiRF serial modem to the MR-C3024 controller on
the RoboNova.
SERVO 02.2008 15
You may want to purchase a special adapter cable from
Lynxmotion (www.lynxmotion.com) called the Sony
Playstation Controller Cable. For $4.95, it is well worth the
money for obtaining the special cable that connects to
the Playstation 2 controller and has convenient connectors
for connecting to your projects. I do not know of any other
place that sells the special connector that the Playstation 2
controller uses.
The MR-C3024 controller
on the RoboNova has two
dedicated serial data ports that
make life really easy for wiring
the BlueSMiRF to the MR-
C3024. One port is for
transmitting serial data, and
the other is for receiving serial
data. Each port has three
wires: a signal wire, +5V, and
GND. The +5V wire (center
pin) is used to power the
BlueSMiRF serial modem. Figure 4 shows a photo of these
ports. Notice the three port labels (ETX, ERX, and LCD)
immediately to the left of the programming port at the
center of the MR-C3024. WARNING: The ETX and ERX
ports are mislabeled on the circuit board. The ERX (serial
receive) port is actually immediately adjacent to the serial
programming port, and the ETX (serial transmit) port is
actually between the ERX and the LCD ports. This is
16 SERVO 02.2008
S
X
2
8
A
C
/
D
P
Vss RB.0
Vdd
RB.1
RB.2
RB.4
RB.3
RB.7
RB.6
RB.5
RC.0
RC.3
RC.2
RC.1
RC.5
RC.6
RC.4
RC.7
RA.3
RA.0
RA.2
RA.1
MCLR
OSC2
OSC1
RTCC
4 MHz
+9V FOR VIBRATION MOTOR POWER
NOT CONNECTED
ACKNOWLEDGE
Vdd (+3V to +5V)
ATTENTION
CLOCK
GROUND
COMMAND
DATA
LYNXMOTION PS2 CONTROLLER ADAPTER
CABLE. AS VIEWED FROM THE FEMALE END
GREEN
BROWN
ORANGE
BLACK
YELLOW
RED/SHIELD
BLUE
VIOLET
N/C
DATA
CMD
ATTN
CLOCK
Vdd
GND
+9V
ACK
4.7 K 4.7 K
10 K
+5V
+5V +5V
R
X
-
I
R
T
S
-
0
G
N
D
T
X
-
0
P
W
R
C
T
S
-
I
M
o
d
e
m
B
l
u
e
S
M
i
R
F
+5V
Figure 2. BlueSMiRF Bluetooth wireless Playstation 2 controller interface.
RX-I
RTS-0
GND
TX-0
PWR
CTS-I
Modem
BlueSMiRF
N/C
RX
N/C
+5V
TX
GND
+5V
GND
TO ETX ON
THE MR-C3024
TO ERX ON
THE MR-C3024
Figure 3. Cable drawing for connecting a BlueSMiRF to the RoboNova.
Figure 4. Serial communication ports on the MR-C3024. Note the
ETX and ERX labels are labeled incorrectly. They should be reversed!
Figure 5. The BlueSMiRF serial modem cable
connected to the ERX and ETX ports.
opposite to what is labeled on the circuit board. The LCD
port is for an LCD display that is available for the robot.
Figure 5 shows the two plugs for the BlueSMiRF cable
attached to these ports. Figure 6 shows the BlueSMiRF
mounted inside the plastic housing on the back of the
RoboNova. The BlueSMiRF serial modem is held in place
with a piece of double-sided foam tape. Double-sided
foam tape works very well.
As you can see, by using the BlueSMiRF serial modem
there isnt a lot of wiring or additional circuits needed to turn
the RoboNova into a truly remote controlled robot. The bulk
of the work is modifying the software on the RoboNova and
establishing a wireless serial link between the RoboNova and
the Playstation 2 controller.
The key to maintaining a good reliable communication
link between the Playstation 2 controller and the
RoboNova is making sure that the two devices are
synchronized so that the data is properly transmitted and
received as expected. The loop structure in the RoboNova
is slower because it has to process the commands and
execute the servo motions. By the time it has finished a
particular command, the time lapse can be from a few
milliseconds to several seconds. The Playstation 2 controller
can cycle through reading the states of the different pins
many orders of magnitude faster. Because both systems
are running independently at different internal loop cycle
rates, there needs to be a way to synchronize the wireless
connection so that the data can be transmitted and
received properly.
There are many methods that can be used to synchro-
nize the wireless link. For example, using the RTS and CTS
features on the BlueSMiRF serial modem. However, I chose
more of a brute force method. When the RoboNova has
completed a particular command and is ready for the next
motion command, it will transmit a four character string,
!PSX. This is to tell the Playstation 2 controller that it is ready
to receive the next motion command. The Playstation 2
controller sits in a loop looking for the !PSX string. When it
finally receives this command, it will then read the current
state of the controllers buttons and joystick positions, and
then transmit the results to the RoboNova with a simple
data checksum. In the meantime, the RoboNova is waiting
until it receives the next command. The four character
string, !PSX, is the synchronization command between the
two systems. You can change this to whatever you want. I
just happened to use this since it is the same synchroniza-
tion method that Jon Williams used in the original
Playstation 2 controller software that helped me figure out
how all this works.
In the main RoboBasic program that the RoboNova
uses, it will have to be modified slightly so that you can use
the BlueSMiRF serial modem. The following is a portion of
the MAIN startup/loop routines from the Overall Template
Program.bas (or you can use the Action_Auto.bas
program) that comes with the RoboNova. This program
will need to be modified to send and receive data instead of
its current configuration to respond to commands from the
IR remote.
================================================
MAIN:
GOSUB robot_voltage
GOSUB robot_tilt
-
IF RR = 0 THEN GOTO MAIN1
ON RR GOTO
MAIN,K1,K2,K3,K4,K5,K6,K7,K8,K9,K10,K11,K12,K13,K14,K15,K
16,K17,K18,K19,K20,K21,K22,K23,K24,K25,K26,K27,K28,K29,K3
0,K31,K32
GOTO main_exit
-
MAIN1:
A = REMOCON(1)
A = A - ID
ON A GOTO
16,K17,K18,K19,K20,K21,K22,K23,K24,K25,K26,K27,K28,K29,K3
0,K31,K32
GOTO MAIN
-
action_proc:
A = RR - 50
ON A GOTO
16,K17,K18,K19,K20,K21,K22,K23,K24,K25,K26,K27,K28,K29,K3
0,K31,K32
RETURN
-
main_exit:
IF RR > 50 THEN RETURN
RR = 0
GOTO MAIN
================================================
Since I wasnt using a gyroscope or monitoring the
battery voltage, I removed the robot_voltage and robot_tilt
call routines from the MAIN program. I removed two of the
three ON n GOTO MAIN,K1, K2,. command sections,
and removed the action_proc subroutine and the main_exit
routine. (I pretty much gutted the main routines.)
The following shows the changes to the MAIN routine
that enable serial communications with the BlueSMiRF serial
modem. The DELAY 5000 is just a five second delay that I
added so that it will allow me some time to turn on the
Playstation 2 controller after I turned on the RoboNova. This
SERVO 02.2008 17
Figure 6. The BlueSMiRF serial modem mounted to
the inside of the RoboNovas back body panel.
can be eliminated or changed. It doesnt matter. The ETX
command sends only one character at a time, and the
character definition must be in a variable. So, to send the
synchronization command !PSX requires eight lines of code.
I am sending the commands at 19200 baud. This was
arbitrarily chosen. I just happened to have configured the
BlueSMiRF serial modems to operate at 19200 baud. You can
choose any other speed. Keep in mind that slower speeds are
more reliable in ensuring proper data transfer, but it also
takes longer to transmit. You may begin to notice a lag time
between transmission and reaction. You will need to
experiment to see which baud rate works best for you and
the RF noise environment that you are in.
================================================
DELAY 5000
MAIN:
A=0
tmp = !
ETX 19200, tmp
tmp = P
ETX 19200, tmp
tmp = S
ETX 19200, tmp
tmp = X
ETX 19200, tmp
ERX 19200, A, main
ERX 19200, tmp, main
IF tmp <> 126 THEN GOTO Main
ON A GOTO
16,K17,K18,K19,K20,K21,K22,K23,K24,K25,K26,K27,K28,K29,K3
0,K31,K32
A=0
GOTO MAIN
================================================
Immediately after sending the !PSX synchronization
command, I read in the next two bytes of data. As with the
ETX, the ERX can only read in one byte of data at a time. The
variable, A, is the variable that contains the actual RoboNova
motion command. This will be a number between 0 and 31.
The Playstation 2 controller will convert the controller
commands into a number between 0 and 31. The second
variable, tmp, that is read from the ERX command is a simple
checksum command. Here, I arbitrarily chose to send a byte
value of 126 as a checksum to verify that the data was sent
properly. There are much better checksum methods that
should be used, but like I said before, I used a brute force
method to make this work. If the checksum value is not 126,
then the RoboNova will consider the data set invalid, and will
begin the loop structure all over again.
I also removed the following lines from earlier in the
program since the robot is no longer using the IR remote:
== Action command check (50 - 82)
IF RR > 50 AND RR < 83 THEN GOTO action_proc
RR = 0
Then I added the following line to the variable declara-
tion set so that it will know that tmp was a byte variable:
DIM tmp AS BYTE
Thats all there is to make the RoboNova program work
with the BlueSMiRF serial modem. The next step is to make
sure that the BlueSMiRF Playstation 2 controller software is
configured to receive and transmit the proper data. The
following is the MAIN routine in the SX28 microcontroller.
MAIN: Wait for !PSX synchronization
char = RX_BYTE
IF char <> ! THEN Main
char = RX_BYTE
IF char <> P THEN Main
char = RX_BYTE
IF char <> S THEN Main
char = RX_BYTE
IF char <> X THEN Main
RoboNova_cmd = 0
Get_Status: returns buttons and joysticks
WAIT_MS 1 give host time to setup
READ_PSX read PSX inputs
if psxStatus = $00 then
PSX_Config
READ_PSX
endif
RoboNova_cmd = 0
if psxJoyRY > 220 then walk forward
RoboNova_cmd = 12
TX_Out RoboNova_cmd
TX_Out 126
goto Main
endif
if psxJoyRY < 25 then walk backwards
RoboNova_cmd = 11
TX_OUT RoboNova_cmd
TX_Out 126
goto Main
endif
if psxJoyRX < 25 then turn left
RoboNova_cmd = 22
TX_OUT RoboNova_cmd
TX_Out 126
goto Main
endif
if psxJoyRX > 220 then turn right
RoboNova_cmd = 24
TX_OUT RoboNova_cmd
TX_Out 126
goto Main
endif
GOTO Main
The beginning part of this routine is an endless loop that
looks for the !PSX synchronization command. Once it
receives this command, the current state of the Playstation 2
controller is read and stored into a set of joystick and
pushbutton variables. Then, the RoboNova command vari-
able RoboNova_cmd is reset to zero. From this point on, the
different joystick positions and button states are converted
into a single byte with a value between 0 and 31. The reason
I chose to do it this way was because the default RoboNova
has 32 preprogrammed motion routines that are mapped to
18 SERVO 02.2008
the numbers 0 to 31 via the ON n GOTO instruc-
tion in the RoboBasic demo program. So to keep
things simple, I kept the same data structure.
Once the RoboNova_cmd is determined, it is
transmitted to the RoboNova via the BlueSMiRF
serial modem. Then the checksum value, 126, is
transmitted to tell the RoboNova that the entire
data set transmission is complete.
The entire programs for the SX28 microcon-
troller for reading the Playstation 2 controller
and commanding the RoboNova robot, along
with the RoboBasic program for the RoboNova
robot can be downloaded from the SERVO
website (www.servomagazine.com). Table 1
shows which Playstation 2 controller button/joy-
stick position controls which built-in RoboNova
motion command. The October 07 issue of
SERVO Magazine has a long discussion on how
to configure the BlueSMiRF serial modem for
automatic connection between two devices, or
you can obtain the configuration instructions
from Sparkfun Electronics. Autoconnect is the
best way to configure both of the BlueSMiRF
serial modems, otherwise you will have to add
some more code to both programs to tell the
BlueSMiRF modems to search for each other
and then to connect to each other.
Table 1 shows a list of the Playstation 2
command sequences and the resulting
RoboNova motion command. These are what I
happened to use from the 32 built-in commands
to choose from. In the Playstation 2 SX/B
program, combination key functions can easily
be programmed in. With the single command
byte that is sent, up to 256 separate commands
can be configured. Modifying the pro-
grams to send two bytes of data can be
used to send over 65 thousand unique
commands. And yes, the Playstation 2
controller has far more command
options than a regular R/C transmitter.
There are a lot of better ways to
do this. This is just what I did, and it
worked well enough for the robot to
compete in the robot Sumo contest at
Robothon. It is also good enough for
the robot to compete in the Robo-One
competitions (www.robo-one.com).
Some of you may be wondering how
my robot survived the competitions. It
did just fine, and still works like it did
before the contest. A few scratches
here and there, but no damage. So,
have fun with this, and let the SERVO
readers know how things go with your
experiments. Many of us really enjoy
learning about what other people are
doing with their humanoid robots. SV
SERVO 02.2008 19
Playstation 2 Controller
RoboNova
Command
Command Description
Right Joystick Up 12 Walk Forward
Right Joystick Down 11 Walk Backward
Right Joystick Right 24 Turn Right
Right Joystick Left 22 Turn Left
Left Joystick Up 21 Forward Tumble
Left Joystick Down 31 Backward Tumble
Left Joystick Right 30 Right Cartwheel
Left Joystick Left 28 Left Cartwheel
Left Arrow 14 Shift Left
Up Arrow 10 Walk Fast
Right Arrow 13 Shift Right
Down Arrow 8 Shoot Pistols
Start Button 2 Raise Arms
Press Down on Left Joystick 19 Backward Standup
Press Down on Right Joystick 25 Forward Standup
Select Button 1 Bow
Square Button 15 Left Attack
Triangle Button 29 Front Attack
Circle Button 20 Right Attack
X Button 7 Flap like a bird
R1 Button 9 Handstand
L1 Button 26 Toggle between sit or stand
R2 Button 4
Sit with raised arms, wait one
second, then stand
L2 Button 3 Sit, wait one second, then stand
Table 1. Playstation 2 wireless controls for the RoboNova robot.
Drobo the Worlds First
Storage Robot
D
ata Robotics has developed Drobo the worlds first
storage robot. Drobo is a striking black desktop
device that connects via any USB 2.0 port on a Mac
or Windows PC. Under its magnetically attached
faceplate are four bays that can accept any brand, size,
speed, or capacity mix of standard 3.5 SATA internal
hard drives (sold separately and available at most
electronics stores).
Drobo has gained a following of enthusiasts including
creative professionals such as graphic designers,
photographers, and video editors, as well as users in the
education sector and small business owners.
Drobo works like any external USB drive and installing
it is as easy as pulling it out of the box and plugging it into
a USB port. However, unlike competitive solutions, Drobo
requires no software, no configuration, and no computer
expertise to start reliably storing important data. In a
ddition, Drobo users benefit from the extensive reliability
features of the storage device, which uses patent-pending
RAID-like proprietary software. Drobo boasts data
redundancy features normally associated with RAID such
as mirroring and data striping, but without the hassle of
RAID. Drobo creates one large pool of protected storage
out of the available drives, enabling users to get rid of
multiple external drives.
Julian Terry, chief architect of Data Robotics and
20-year-veteran of the storage industry, said he was
inspired to create Drobo after losing irreplaceable
personal data from a system crash at home. After
inventing enterprise storage technologies and building
entire data centers, I understand how to store and protect
data better than most, Barrall said. When I investigated
what solutions were out there, the only options were
managing a growing number of external hard drives or
managing RAID. Thats when I realized there needed to be
an easier solution, and I began the development of Drobo
four years ago.
Drobo is distributed in Sweden by FixIT and is available
immediately.
For further information, please contact:
Sub-50 Cent Eight-Bit Flash
Microcontroller Family
S
TMicroelectronics now offers a new ultra-low-cost
series of Flash microcontrollers (MCUs) intended for
highly cost-sensitive products such as electric bicycles
(ebikes), air conditioning, small appliances, and
sensors. The ST7FOX series includes devices with from
2KB to 8KB of on-chip Flash memory, and with eight to
32 pins, all of which are priced below the critical 50 cent
($0.50) threshold.
Part of the well-established ST7 family of MCUs built
on an industry-standard eight-bit architecture, the new
devices achieve their low unit cost with a simplified feature
set, while maintaining the benefits of a robust and proven
technology. ST7FOX operates with +5V (10%) power
supply, and includes a relaxed RC oscillator which can be
calibrated within the application.
The CISC (complex instruction set computer)
architecture is a significant advantage for the production
of compact code matching the small memory footprint of
the familys devices, with nested interrupt management
and a variety of addressing modes in the core architecture
simplifying high level language coding.
ST7FOX is designed to satisfy the increasing
needs in the industry for reliable low-end and very
low-cost MCUs for use in cost-critical applications.
However, its upgrade path to the easy-to-use ST7Lite
family itself a cost-effective solution also enables
manufacturers to build scalable product families to
address broader markets.
The very low pricing enables designers to implement
a microcontroller solution to replace electromechanical or
logic-based control, and for the same cost as ROM devices,
this Flash-based family will ease inventory management
and simplify the supply-chain. A free integrated
development environment (IDE) including a C compiler
minimizes costs for developers.
Samples of ST7FOX are available now in eight-pin DIP
or SO-8 packaged devices with 2KB of Flash, and 32-pin
DIP or LQFP devices with 4KB of Flash. All are priced at less
than $0.50 in large quantities.
New Products
ACCESSORIES CONTROLLERS & PROCESSORS
N
N
E
E
W
W
P
P
R
R
O
O
D
D
U
U
C
C
T
T
S
S
20 SERVO 02.2008
Website: www.drobo.se
Data Robotics, Inc.
Website: www.st.com
STMicroelectronics
FEB08NewProd.qxd 1/10/2008 1:07 PM Page 20
Rugged, Low-Cost Motorized
Robot Platform
T
he new wheeled robot
platform from Electronix
Express is made from
0.1 inch industrial-grade
aluminum. It features
two seven-inch octagonal
plates separated by four pre-
drilled 2.5 inch riser brackets.
You can use the bracket holes to
mount sensors and other accessories. The plates have
grommeted holes for passing cable, and are thick and
strong enough to drill and tap additional mounting holes.
The platform includes an industrial-grade ball-caster
and two reversible 12 VDC gear-head motors with
neoprene foam wheels. It comes assembled, and includes
hex keys for the socket-head chassis-screws and the
set-screws on the wheels. A dual H-bridge is available for
separate purchase. The platform is catalog number
01BRPL and sells for $109.95 each (quantity discounts
are available).
Digitizing Probe
T
ormach offers a touch-trigger CNC Digitizing Probe
that mounts in a machine spindle for fast and accurate
electronic measurement of parts and fixtures.
The probe features a robust detachable ruby tipped
stylus and will connect directly to the Tormach PCNC 1100
auxiliary socket. The probe enables accurate measurement
of parts and sets up jobs quickly. Probe features include:
Capable of generating a point cloud for reverse
engineering and repair applications.
Works with Tormach machine control software to
automate many probing operations.
Compatible with standard Tormach Tooling System
geometry, allowing setup consistent with TTS tool
changes.
By mounting the probe in a machine spindle and
connecting it to the control computer with the proper
software, the machine can be used for measuring instead
of cutting.
The measurements can be used to set work offsets,
measure features, or to generate a 3D point cloud
describing a complex surface.
The Tormach machine control software has interactive
screens to automate standard probing operations such as
finding the center of a bore, finding a plane surface, or
finding the corner of a vise jaw or workpiece.
The probe interfaces directly with the Tormach
PCNC 1100. It can be used with many
other machines provided the electrical interface is
configured properly.
Detailed information on using the probe is
available in Chapter 8 of the PCNC 1100 manual. It is
available online at www.tormach.com/documents.htm
#manual.
The Digitizing Probe (PN 30668) ships in a sturdy
carrying case with an alignment tool and shielded cable.
It connects directly to the PCNC 1100 auxiliary socket.
The included stylus is a 0.1575 (4 mm) diameter ruby
tipped sphere.
The stylus with the ruby tipped sphere is very accurate
and wear resistant, but not inexpensive. Tormach offers a
less expensive steel stylus training kit with a breakaway
section called the Practice Probe Tip (PN 30695) that
is helpful when learning how to operate the probe.
The Probe Wrench Set (PN 30673) is necessary to
swap probes.
Specifications
Probe Dimensions (with tip): 1.6 diameter x 7.0 length
(40 mm x 178 mm).
Cable Length: 120 (3 m) overall, first 40 (1 m) are
shielded.
Probe Weight: 1.4 lbs. (0.6 kg).
Shipping Weight: 4 lbs. (1.8 kg).
TOOLS & TEST EQUIPMENT
CONTROLLERS & PROCESSORS
Is your product innovative, less expensive, more functional,
or just plain cool? If you have a new product that you
would like us to run in our New Products section, please
email a short description (300-500 words) and a photo of
your product to:
newproducts@servomagazine.com
Show Us What Youve Got!
365 Blair Rd.
Avenel, NJ 07001
Tel: 8009722225 or
7323818020 (in NJ)
Website: www.elexp.com
Electronix
Express
SERVO 02.2008 21
204 Moravian Valley Rd., Ste. N
Waunakee, WI 53597
6088498381 Fax: 2098854534
Website: www.TORMACH.com
TORMACH
FEB08NewProd.qxd 1/10/2008 1:07 PM Page 21
Featured This Month:
Features
22 The Holy Grail of Combat
Robotics Usable Melty
Brain: Part 1
by Kevin Berry
25 MANUFACTURING:
Metal Origami by Russ Barrow
Events
26 Nov/Dec 2007 Results and
Feb/Mar 2008 Upcoming
Events
27 HORD Fall 2007 Report
by Chris Olin
ROBOT PROFILE Top
Ranked Robot This Month:
29 Bounty Hunter by Kevin Berry
22 SERVO 02.2008
O
ne of the classic trade-offs in
combat design is armor vs.
platform vs. weapon weight. A
truly efficient, traditional spinning
weapon bot might devote
2530% of total weight to the
weapon. The next step up in
efficiency is a full body spinner,
where weapon and armor are
combined, and 50% or more of
the weight gets applied towards
offensive capability.
A method sometimes used to
get 100% offense is the thwack-
bot, where the entire bot is spun
and a protruding hammer or spike
imparts the built-up momentum
into the opponent. The only
problem with a thwack style bot is
that it basically must spin in place,
hoping the opponent moves into
its kill zone.
Obviously, the optimum
solution is to spin the whole bot
continuously, while also chasing
down the opponent. To do this, a
very complex driving scheme
emerges, known as translational
drift, or in the vernacular, melty
brain. This means inputting slight
speed changes to one set of
wheels, resulting in a drifting move-
ment rather than spinning in place.
Few, if any, drivers have done
this successfully by hand.
Attempts to automate this
function exhibit much technical
creativity, and some limited
success. Searching the various
combat forums reveals periodic
by Kevin Berry
Usable Melty Brain
Part 1: Profiles of (Nearly)
Successful Attempts
THE HOLY GRAIL
OF COMBAT
ROB TICS
CombatZone.qxd 1/10/2008 10:03 AM Page 22
SERVO 02.2008 23
and interesting threads on the
topic, beginning back in 2001, and
continuing up to today. There is also
a Wikipedia article of some interest
on the topic.
Blade Runner
The first credited attempt at
building a melty brain bot belongs to
Team Carnivore, which competed at
the first, Long Beach BattleBots
event, and many others. Ilya Polyakov
contributed his remembrances on
the naming of the technology, and
what drove it.
My middleweight Carnivore
got beaten really bad by Christian
Carlbergs thwack-bot Knee Breaker
at the Long Beach BattleBots. The
thwack-bot design intrigued me and
really got stuck in my head, perhaps
reinforced by flashbacks of the
nasty and unnecessary loss to Knee
Breaker. The design is brutally
simple and perhaps the most effi-
cient conversion of stored electrical
energy to kinetic impact energy
while maintaining reliability,
minimizing moving parts and
maximizing defensive capabilities.Only
one drawback mobility. Because
the robot relies on a tank style drive
train, the robot can only spin in one
spot. Blade Runner 1 was designed
to take advantage of this system.
The concept was to use this robot in
conventional pushy or hand to
hand combat while having the
ability to spin up and attack the
opponent like a full body spinner.
The melty brain system was given
that name as a result of a late night
conversation with one of my
teammates who reported his brain
melting when I tried to explain the
math behind it.
Next months article will discuss
the technical aspects of the transla-
tional drift implementation in all four
featured bots. For this issue, well
focus on the more traditional config-
uration items on each machine. Ilya
gives a short history of Blade Runner:
BR1 was driven mostly as a
pushy bot with some minor side-
ways thwacking. Blade Runner
2 was an all composite version
of BR1 powered by two four
inch Magmotors. This crammed
a 8HP+ drive train into a mid-
dleweight. The bot was amaz-
ingly powerful and destructive,
but keeping the wheels from
losing traction during normal
driving was a feat. In the end,
the all composite nature of
this robot was its un-doing.
Lightweight armor and struc-
ture was critical for meeting the
120 lb weight limit with some 80 lb
of just motors and batteries. The
armor and structure were com-
bined into a single monocoque
Kevlar shell. The all composite
construction entailed fabricating
everything out of Kevlar, including
bearing mounts, motor mounts,
and battery mounts. The Kevlar
works great as armor where high
tolerances are okay, however, trying
to line up motor mount holes and
bearing mounts in Kevlar was too
much. The project ran weeks late
resulting in the final robot assembly
taking place in the back of a
Suburban on the drive between San
Jose and Treasure Island! We did
not have time to implement the
melty brain system. Blade Runner
competed unfinished in both sea-
son 5 and 6 of BattleBots. Season 5
had a few wins but a forgotten clip
resulted in a lost wheel and a loss.
Season 6 was ended by an amazing
fight against Hazard where one of
the hits penetrated 3/8 thick lami-
nated Kevlar armor and cut through
a Magmotor sick!
BotRank lists Blade Runner as
5-5 across all fights, from BattleBots
1.0 through 5.0.
CycloneBot
Next up in our
melty brain hit list is
CycloneBot from CM
Robotics. This is a 220
pound Heavyweight
which fully implements
the translational drift
strategy with a sophisticated comput-
er system. The backbone of the bot is
an oblong, one piece shell, made
from 1/2 welded titanium. The bot
features impressive craftsmanship,
along with a proprietary system
called TauntWare which allows
real-time, marquee style messages
to be displayed on the side of the
rotating bot during a fight. The team
has termed their drive scheme
Cyclone Drive, and has garnered
much interest among the computer
and electronics industry for their
implementation.
Michael Worry from the team
describes the platform and some
interesting problems: Weve gone
through a few different iterations
on the system. The melty brain is
continuously accelerating and
braking the motors which is
very hard on the motor controllers,
batteries, and wiring.
Today, we use Odyssey 12V
SLAs, two in series and run a 24V
bus into two MC1-HVs. We find we
need to use the dry cell SLAs, as
any sort of liquid in the batteries
causes battery failures when
spinning at 300 gs.
Blade Runner
CycloneBot
24 SERVO 02.2008
Weve actually taken batteries
out of the robot that are visibly
bulged on one side. We started with
NiCds and NiMHs, but kept burning
them up due to insufficient power
density.
The MC1-HVs are fed from a
custom board that CM Robotics built
that provides isolation between the
digital and power electronics. The
digital electronics are a ruggedized
NIOS development board from
Altera.
On motors, the original
CycloneBot used two Etek motors
which worked awesomely. However,
we found the power density of the
system was limited by the power
from the batteries rather than the
power output of the motors. Plus,
we wanted to go to a much smaller
chassis to permit thicker armor.
We changed to four longmag motors
two on each side into a Whyachi
gearbox. These longmags have
worked reasonably well, though
we have problems with the impact
forces dislodging the magnets in
the casing.
We use carefree tires,
machined down to our desired diam-
eter. Armor and chassis are all titani-
um. Blades are hardened tool steel.
CycloneBots BotRank record is
11-20, fighting in a wide variety
of the sports most prestigious
events since 2003. Their website is
www.maccanikill.com
Melty B
A recent addition to the melty
fleet is Melty B, an antweight
meeting with some success in the
box. Team SpamButchers builder,
Rich Olson, built many test platforms
and a 1.0 version before hitting suc-
cess, winning Robothons antweight
division in September 2007.
Rich Olson describes Melty Bs
guts. It uses Ever Motors ERS-365S-
3033s. Theyre similar to a Speed
400 except theyre a bit shorter
and lighter. Theyre rated at 6V with
16,340 no-load RPM. Im running
them at 7.4V and getting about
8,000 RPM (theres a lot of load
and theyre only on about 75% of
the time due to the melty brain
stuff). The battery is a 7.4V 910 mA
Thunderpower LiPo.
For motor controllers in the
most recent (1 lb) version, I actually
used Darlington drivers (ST Powers
BU941ZT). These are inexpensive
(about $2.50 each) and I was able to
connect them directly to the micro-
controller. I am using one Darlington
driver per motor. Since the drivers are
only on/off (no reverse) the robot
is only translational drift it has no
normal drive mode.
The shell is a custom cut 7
aluminum (6061) pipe segment
(1/8th thick). I got this from www.
onlinemetals.com for under $10!
The knockers are titanium and
Delrin. The Delrin held up pretty well
but did eventually crack in one of
my final matches. Most of the
electronics in the bot are connected
using wire wrapping, and the top
and bottom are made of .03 thick
garolite (fiberglass).
A nice historical build page
including the total software load and
videos is available at www.spam
butcher.com.
BotRank shows that Melty B 2.0
went 4-0 at Robothon 2007.
Scary-Go-Round
The last in our line-up of
bots using this drive scheme
is Scary-Go-Round, a 30 lb
Featherweight from Dales
Homemade Robots. Dale
Heatherington, a recognized
innovator and master craftsman,
got interested in the translation-
al drift schema, after following
CycloneBots successes. He started,
as did Rich, with models and
prototypes before turning out a
finely crafted machine.
Three modular motor/gearbox
units were built on a CNC mill, pow-
ered by Team Orion 13 turn R/C car
motors, driving 4 wheels. A 23:1
ratio in the gearbox ensures compli-
ance with Robot Battles 20 fps
maximum for spinning weapons. The
unique three legged Y configura-
tion puts maximum mass out at the
perimeter to maximize Moment Of
Inertia. After experimenting with
spring bumpers intended to launch
the opponent (but, which proved to
be more effective launching Scary),
he turned to wedges. To get around
the inherent problem of zero
clearance wedges (hanging up and
launching on arena floor bumps), he
came up with a smart wedge that
retracts upon hitting an obstacle.
Robot Battles features 1/4
aluminum bars all over the platform.
Shop trials showed that Scary could-
nt translate over the bars, so Dale
was forced to come up with another
innovation: an automatic conversion
from melty drive to standard tank
drive. A rotatable skid falls under
a wheel when the bot is spun in
reverse, converting it to tank drive.
Dale has posted a superb build
report with photos and videos, at his
site: www.wa4dsy.net/botblog.
Unfortunately, like so many first
time bots, Scary-Go-Round fell prey
to that common bugaboo a loose
Melty B
Scary-Go-Round
Material and photos were contributed by
Ilya Polyakov, Team Carnivore; Rich Olson,
Team Spambutcher; Michael Worry,
CM Robotics; and Dale Heatherington,
Dales Homemade Robots
SERVO 02.2008 25
W
eight is perhaps the single most
difficult challenge in the devel-
opment of any robotic system. While
this design aspect may not seem the
top concern for your robotic system, it
will influence every part of it, from
materials used, to mobility, speed, and
power. In addition, for each robotic
design, there is the builders need to
design something unique or artistic.
For me, designing with minimal
weight requirements is simply part of
the creation process. To accomplish
this, it is important to understand the
simple concept of a form.
In combat robotics, the form or
exterior design is paramount to the
defensive and offensive strategy
employed, but in general the exterior
design of any robot must match the
requirements of the task. A brilliant
electrical design or application can
be lost in an ordinary black box.
Formed shapes and complex angles
can provide not only a more pleasing
or natural shape, but they can also
dramatically increase material rigidity
and deformation resistance. Building
such designs does require more
planning and the willingness to
potentially abandon familiar CAD
programs for initial development.
Although each builder should employ
techniques that they are comfortable
with, I will discuss several techniques
I have used to convert flat sheet
metal into dynamic sculptures.
Often, when many people build
a robot or machine, they think about
familiar dimensional characteristics
such as a top, bottom, right side, left
side, front, and back. Then they take
their measurements and cut their
materials to match these dimensions.
Joining the material is often an
afterthought accomplished with
welding, fasteners, and too often
glue. Unfortunately, these solutions
often result in a joint that is not as
strong as the material used or severe-
ly weakens the material through
irregular heat zones, dissimilar
metals, cuts, or holes at the joints.
Another issue is that most
designs use flat planar lengths of the
material that do not provide strength
in multiple axis. This often causes a
lack of rigidity across the material
requiring weight inducing reinforce-
ment. Simply creating an edge in the
material will significantly improve
rigidity along the edge. To demon-
strate this, imagine a single sheet of
paper. Unfolded, the paper will sag
regardless of where it is held, but
fold the paper and create a crease,
and the sheet will now have a rigid
span along the crease. This edge is
important since it demonstrates the
inherent advantage of forming and
the continuity of the material that
does not require fasteners or other
joining techniques (Photo 1).
Angular bends are not the only
way to improve material rigidity and
multiple axis strength. Rounded
forms can also create very strong
shapes such as domes and cylinders.
Although a dome cannot easily be
demonstrated with paper, most
plastics and metals can be stretched
along the interior of the material
using a hammer, heat, and/or rolling
(using a tool such as an English
wheel). A half dome is another
shape that can be formed by using
half of a circle or ellipse. By bending
the shape perpendicular to the half
cut, an even curve in the material will
produce a domed shape that directs
energy away from the center of the
form toward the edges (Photo 2).
Now that the ideas of folding
the material have been understood,
the next step is to see how a design
could be created in this manner. If
you are more comfortable creating
each component separately, scale
the parts and create them on paper.
Now, with a little tape, join their
common edges. Slowly, the connect-
ed pieces should take on a single
piece form. One note: The connected
edges should match the places
where the most strength is required.
There are CAD tools that can create
these forms as well, however, paper
MANUFACTURING:
Metal rigami
by Russ Barrow
wire causing it to lose at
Robot Battles.
This article provides just a peek
into the history of attempts at build-
ing translational drift driven bots. In
the interest of magazine sales, Ive
withheld all the juicy technical details
for Part 2, Looking Under The Hood:
The Technology Of Melty Brain. SV
PHOTO 1
PHOTO 2
26 SERVO 02.2008
Nov. 12-Dec. 16, 2007
R
oaming Robots held their Winter
Tour Grand Finals in Nottingham,
England at the Harvey Hadden Sports
Complex on December 1st. Go
to www.roamingrobots.co.uk for
more information. Results are
as follows:
Heavyweights
1st: Iron Awe
is a cheap and easy way to prove out
an idea (Photos 3 and 4).
Next, we need to form the flat
sheet based on the type of material
used. An industrial brake press is a
handy tool that many of us will not
have access to, however, some
materials can be formed by hand or
a common vise with the use of heat
or brute force.
Angular bends work well with a
vise and a block material that evenly
distributes the force you are
applying. The block can be metal,
wood, or plastic, but typically is flat
and more rigid than the material
being bent. Rounded forms (such as
the half doom example above) are
best shaped by holding down a
corner or mid section and applying
force to the end of the piece, there-
fore curving the material as needed.
As an example, I have placed a
metal sheet in an expansion joint
between two sections of concrete,
and then used the brute force of my
weight by leaning or stepping on the
material to get the shape I wanted
(Photo 5).
Plastics and some metals will
respond well to being bent without
heat, but some materials do not
easily stretch and will actually fatigue
if bent. Titanium and aluminum are
materials that will greatly benefit
from a large radius rolled edge or
sharp crease of less than 30 degrees.
If a tighter bend is necessary with
these materials, or you cannot easily
brute force thicker metals or thermo
plastics, then the application of heat
may be necessary.
Heat can be crated from many
sources, including a blow torch,
propane gas torch, or even a small
cylinder of Mapp Gas. The key here is
to let the heat do the work. Securely
mount the material and use the hottest
part of the flame (generally the blue tip
of the flame) to heat the material.
Titanium and steel will become a
dark red when they are ready to be
formed. Aluminum and plastics are
more difficult to determine when
the best forming temperature is
achieved, so keep continual force on
the material until it begins to soften
and bend. For lengthy angular
edges, keep the torch moving over
the material along the edge, and use
a block to apply consistent force.
Heat can also be used to transfer
a shape or form to the sheet of materi-
al. To create a radius bend, you can
secure the material and use pipe or
tubing placed against the sheet with
the needed radius to create a roll in the
material. Once again, keep the torch
moving back and forth along the bend
edge to keep the radius consistent.
Transferring other shapes can be
done by placing the material over the
form and allowing the heat to sag
the material to match the form using
nothing more than gravity or light
force from a hammer or pliers
(Photos 6 and 7).
It will take some time to master
the techniques above, but with a little
patience, you can make something
that is both stronger and more aesthet-
ically pleasing than a joined box. SV
EVENTS
Results and Upcoming Events
PHOTO 3 PHOTO 4
PHOTO 5 PHOTO 6 PHOTO 7
SERVO 02.2008 27
5, Team Iron Awe, Rob and Gilbert
Grimm; 2nd: Kan Opener, Team
Kan Opener, Andy and Tom
Kane; 3rd: Tilly Ewe 2, Team Tilly,
Shane Lale.
Featherweights 1st: Beauty 2,
Team Beast, John and John Jr
Lear; 2nd: Mini Mighty Mouse,
Team Mouse, Trevor Wright; 3rd:
Skink, Team Windbags, Mike
Hamilton-Macy.
B
otsIQ held their Massachusetts
Regi onal
Competition in
Boston on
December 1st.
Twenty five
teams partici-
pated. Go to
www.botsiq.
org for more information.
Upcoming Events for
February-March 2008
W
AR NW
Mo d e l
Hobby Expo
2008 will be
presented by
Western Allied
Robotics in
Seattle, WA on 2/9/2008. Go to
www.westernalliedrobotics.com/
for more information. The event will
be held at the Monroe County
Fairgrounds (near Seattle); www.
nwmodelhobbyexpo.com/Directi
ons.html. The event will be
held from 11:30am-6:00pm, safety
inspection from 9:00am-11:00am. If
a lot of robots register, they may
start safety and fights earlier.
Classes: 1, 3, 12, and 30* pound
robots (*30 lb spinning-weapons not
allowed). Format: Double Elimination
or Round Robin (RFL Rules). No
ICE or open flames. Entry Fee: $40
for first 30lb or 12lb robot. $25 for
first 3lb or 1lb robot. Additional
robots are half price. Special entry
fee considerations for builders
who are under 18. Arena: 12 x 12
with 18 x 18 pit in one corner
surrounded by a 2 high wall making
it very difficult to accidentally drive
into it.
R
obots Live will present an event
at the Hermitage Leisure Centre,
Whitwick, Leicestershire on February
2nd and 3rd, 2008. Go to www.
robotslive.co.uk for more
information. For the first event of the
year, Robots Live heads back to
their home town of Whitwick in
Leicestershire. Not only is this their
home town, but also their birthday!
Come and celebrate two years of
Robots Live!
M
otorama 2008 will be present-
ed by North East Robotics
Club, Inc., in Harrisburg, PA, from
2/15/2008 through 2/17/2008. Go
to www.nerc.us/ for more
information. This will be an 150 g to
30 lb combat event. Fairies and Ants
fight in 8 box on Friday; Beetles
through Featherweights fight in
16x16 box on Saturday and Sunday.
All completed forms and entry fees
must be received by 1/2/07. This is
going to be another awesome event
at the Farm Show Complex!
R
oaming Robots will present
an event on February 17th, 2008
at The Metro Dome in Barnsley, UK
and another on March 22nd
and 23rd, 2008 at Colchester
Leisure World in Colchester, UK. Go
to www.roamingrobots.co.uk for
more information.
C
entral Illinois Bot Brawl 2008 will
be presented by Central Illinois
Robotics Club in Peoria, IL on March
29th, 2008. Go to http://circ.
mtco.com for more information.
Categories will be: RC Combat (1 lb
Ants only), Autonomous Sumo (3 kg,
500 g, LEGO), Line Following, Line
Maze. Entry fee is $7 per entry if you
pre-register, $10 per entry for walk-
ins. Free admission for spectators.
B
otsIQ will hold a Regional
Competition on March 28th and
29th, 2008 in Pittsburg, PA. Go
to www.botsiq.org for more
information. SV
O
n Saturday, November 3rd, the
Ohio Robotics Club (ORC) in
association with the Robot Fighting
League (RFL) held their 6th insect
weight combat robot event, House
of Robotic Destruction, Fall 2007
(HORD) at the Cuyahoga Valley
Career Center (CVCC).
This event helped kick off
CVCCs new robotics program.
Students in this program will be
building ant weight robots for the
next ORC/CVCC event to be held
next spring.
Eleven teams brought a total of
H RD FALL 2007 REPORT
by Chris Olin
20 robots to compete in three
weight classes; 150 g Flea Weight, 1
lb Ant Weight, and 3 lb Beetle
Weight. Teams came from across
Ohio, as well as Pennsylvania,
Indiana, and most notably, Team
Mechanical Advantage flew in all the
way from Kansas.
Action started with three Flea
Weight robots fighting a double
round robin tournament. Little Buzz,
driven by Richard Kelley of Boiling
Spring, PA, dominated the class
winning four straight matches. 0-2,
driven by Evan Gandola of Omsted
Township, OH, exceeded expecta-
tions by finishing 2-2; while I.R.D.
(Improvised Robotic Device), driven
by Chris Olin of Perry Township, OH,
failed to win a single match, largely
due to a lack of reliable batteries.
In Ant Weight action,
eight robots fought through a dou-
ble elimination tournament. Dusty
the Evil Dustpan, driven by Jeff Gier
of Kansas City, KS, managed to come
back from an early defeat and fight
his way to the finals.
Heman I, driven by Evan
Gandola muscled his
way through the winners bracket to
the finals, but not even the power of
Greyskull could stand against the
determined dustpan. Dusty pitted
Heman and claimed first prize.
Moving on to the Beetle
Weights, nine robots clashed in a
double elimination tournament. D2,
driven by David Timothy of Brook
Park, OH, tore a path of destruction
through the winners bracket all the
way to the finals. Meanwhile, The
Box, driven by Richard Kelley, after
losing in the second round pushed
back through the losers bracket to
the finals. But, his comeback was
halted by the 1 lb spinning drum of
D2. D2 K.O.s The Box and took home
the first prize honors.
In rumble action, Heman I
prevailed over four other contenders
to win the Ant Weight rumble, while
Sweaver, driven by Greg Shay of
Garrettsville, OH, beat out five
other robots to win the Beetle
Weight rumble.
Prizes and other considerations
were provided by: Dimension
Engineering (www.dimensionengi
neering.com); Micro Bot Parts
(www.microbotparts.com);
SERVO Magazine (www.servo
magazine.com); and Cuyahoga
Valley Career Center (www.
cvccworks.com).
ORCs next event will be
April 19th at Cuyahoga Valley
Career Center. See www.
ohiorobotclub.org for more
details. SV
28 SERVO 02.2008
Left to right: Richard Kelley, Team Kelley,
Boiling Spring, PA, driver of Little Buzz. Evan
Gandola, Team Probotics, Omsted Township,
OH, driver of 0-2.
D2 Little Buzz
Dusty The Evil Dustpan
Left to right: Evan Gandola, Team Probotics,
Omsted Township, OH, driver of Heman I. Jeff
Gier, Team Mechanical Advantage, Kansas
City, KS, driver of Dusty the Evil Dustpan.
Greg Shay, Team Fishneck, Garrettsville, OH,
driver of Sweaver. Richard Kelley, Team Kelley,
Boiling Spring, PA, driver of The Box. David
Timothy, Team D2, of Brook Park, OH, driver of D2.
Evan Gandola, Team Probotics, Omsted
Township, OH. Jeff Gier, Team Mechanical
Advantage, Kansas City, KS. Greg Shay,
Team Fishneck, Garrettsville, OH. Richard
Kelley, Team Kelley, Boiling Spring, PA.
David Timothy, Team D2, Brook Park, OH.
B
ounty Hunter has competed in
Motorama 2007 and Franklin
Institute 2007. Details are listed below:
Frame: Internal frame work is
aluminum
Base plate: 3/16 aluminum with
lots of large holes to save weight
Drive train: Four wheel drive, Black
and Decker 9.6V drill motors
Voltage: 12.0 volts
Wheels: 4 x 0.75 Colson
Configuration: Wheels narrower in
front than rear to keep inside armor
Drive ESC: Two IFI Victors
Drive batteries: Saft 3.2 Ah
Weapon: Flipper
Weapon power: Co
2
system, full
pressure, 850 psi direct
Weapon control: Four way valve
from McMaster-Carr is fired by
Team Delta R/C switch
Armor: All titanium, very thin
Future plans: Needs more power,
larger ram, go to two wheel drive
set-up to save weight
Design philosophy: Flippers win SV
Photos and information are courtesy of
Jerrk Clarkin and Brian Benson. All fight statistics
are courtesy of BotRank (www.botrank.com) as
of December 11, 2007. Event attendance data is
courtesy of The Builders Database (www.
buildersdb.com) as of December 11, 2007.
ROBOT PR FILE
by Kevin Berry
TOP RANKED ROBOT THIS MONTH
Weight
Class
Bot Win/Loss
Weight
Class
Bot Win/Loss
150g VD 26/7 150g Micro Drive 10/2
1 lb
Dark
Pounder
44/5 1 lb
Dark
Pounder
28/3
1 kg Roadbug 24/10 1 kg Roadbug 14/5
3 lb 3pd 48/21 3 lb Limblifter 15/3
6 lb G.I.R. 14/2 6 lb G.I.R. 11/2
12 lb Solaris 42/12 12 lb
Rants
Pants
16/3
15 lb Humdinger 26/4 15 lb Humdinger 26/4
30 lb Helios 31/6 30 lb Billy Bob 8/0
30 lb
(sport)
Bounty
Hunter
9/1
30 lb
(sport)
Bounty
Hunter
9/1
60 lb
Wedge of
Doom
43/5 60 lb
Son of
Whacky
Compass
8/0
120 lb
Devil's
Plunger
53/15 120 lb Touro 9/2
220 lb
Sewer
Snake
35/9 220 lb Brutality 8/2
340 lb
SHOVELH
EAD
39/15 340 lb
Psychotic
Reaction
4/1
390 lb MidEvil 31/9 390 lb MidEvil 3/0
Top Ranked Combat Bots
Rankings as of November 13, 2007
Historical Ranking is calculated by
perfomance at all events known to
BotRank
Current Ranking is calculated by
performance at all known events,
using data from the last 18 months
History Score Ranking
Historical Ranking: #1
Weight Class: 30 lb Sportsman's Class
Team: Hammertime
Builder: Jerry Clarkin
Location: Malvern, PA
BotRank Data Total Fights Wins Losses
Lifetime History 10 9 1
Current Record 10 9 1
Events 2
Bounty Hunter Currently Ranked #1
New
Section!
SERVO 02.2008 29
Know of any robot competitions Ive missed? Is your
local school or robot group planning a contest? Send an
email to steve@ncc.com and tell me about it. Be sure to
include the date and location of your contest. If you have a
website with contest info, send along the URL as well, so we
can tell everyone else about it.
For last-minute updates and changes, you can always
find the most recent version of the Robot Competition FAQ
at Robots.net: http://robots.net/rcfaq.html
R. Steven Rainwater
F Fe eb br ru ua ar r y y
24-28 APEC Micromouse Contest
Austin Convention Center, Austin, TX
Amazingly fast little autonomous robot critters race
to solve a maze in this competition. If youve never
seen one of these events, go see this one. You
wont believe how fast these things are.
www.apec-conf.org
28- Pragyan
Mar 2 National Institute of Technology, Trichy, India
Events include standard Micromouse and Sym-Bot,
a contest in which a remote controlled robot must
guide an autonomous robot to the starting line
of a course then the autonomous robot must
complete the course by itself.
www.pragyan.org/08/home/events/
robovigyan
M Ma ar r c ch h
7-8 AMD Jerry Sanders Creative Design Contest
University of Illinois at Urbana-Champaign, IL
Check the website for the details of this years
contest.
http://dc.cen.uiuc.edu
8 Fort Collins Robot Fire Fighting Challenge
Discovery Science Center, Fort Collins, CO
This is a regional for the Trinity College Fire Fighting
Robot contest. Autonomous robots must locate
and extinguish a flame in a scale model of a home.
www.strout.net/fcrf fc
15-16 Manitoba Robot Games
Tec Voc High School, Winnipeg, Manitoba, Canada
Included in this competition are a mix of events
for autonomous and remote-controlled robots
including Japanese style mini-Sumo, Western style
Sumo, a robot Mini-Tractor Pull, Super Scramble,
line-following, and the Robo-Critters contest for kids.
www.scmb.mb.ca
29 CIRC Central Illinois Bot Brawl
Lakeview Museum, Peoria, IL
This event includes RC combat, autonomous Sumo,
line-following, line maze. Autonomous and remote-
control robots.
http://circ.mtco.com
30 Boonshoft Museum Robot Rumble
Boonshoft Museum, Dayton, OH
This event includes robot building and competition.
www.boonshoftmuseum.org
TBA Penn State Abington Mini Grand Challenge
Penn State Abington, Abington, PA
This event includes outdoor autonomous mobile
robot navigation.
http://www.ecsel.psu.edu/~avanzato/robots/
contests/outdoor/contest05.htm
TBA DPRG RoboRama
Dallas, TX
This event includes quick trip: an entry-level event
where the robot moves from A to B and back;
line-following: the robot must follow a line on the
floor; T-time: the robot moves through three points
on a T-shaped course; and can can: the robot must
locate and retrieve empty soda cans.
www.dprg.org/competitions
A Ap pr ri i l l
26 RoboFest
Lawrence Technological University, Southfield, MI
This event includes game competition
two autonomous robots work together. Also robot
exhibition, RoboSumo, RoboFashion show, and
mini urban robot challenge.
http://robofest.net
Send updates, new listings, corrections, complaints, and suggestions to: steve@ncc.com or FAX 972-404-0269
30 SERVO 02.2008
Events.qxd 1/10/2008 12:52 PM Page 30
SERVO 02.2008 31
I
ve tried various simple control
projects before and sometimes
the devices themselves were not
spectacularly successful. A disastrously
blind line-following robot that
challenged both my mechanical skills
and my patience comes to mind. Still,
my students found these projects
interesting and motivating, much
more so than the usual static sort of
class project.
When I ran across Vellemans
K8055 USB Experiment Interface Board
kit, I realized it could form the basis of
a simple, PC-based control system. I
had used various embedded micro-
controllers before, but the USB/PC
combination had some definite
advantages. We wouldnt need a cross-
compiler, there were no serial links to
(mis)configure, and we could use the
FIGURE 1. RoboCooler Block Diagram. This
shows the major components of the cooling
and heating system, including the PC, the K8055
controller, the interface board, the sensor, and
the output devices. Devices shown within the
large rectangle are integral to the K8055 board;
all other components are externally connected.
I teach introductory programming classes at a local community college
and Im always looking for an involving project for my students. One of the
problems with entry-level programming classes is that many students find the
example programs they construct boring and somewhat removed from the real world. Its hard for
anyone myself included to get too excited about coding a bubble sort, for instance.
Reed.qxd 1/10/2008 10:37 AM Page 31
32 SERVO 02.2008
Microsoft Visual C++ compiler installed
in the classroom lab with which the stu-
dents were already becoming familiar.
This article recounts how we
created and tested the RoboCooler,
essentially a model heating and air
conditioning system based around a
disused Igloo cooler, the Velleman
interface board, and a handful of sur-
plus parts. While the control algorithm
we implemented is elementary, readers
may find some of the techniques and
technology useful in more sophisticat-
ed systems of their own devising.
System Overview
Figure 1 provides an overview of
the system. The Velleman board con-
nects via USB 1.0 to a PC running
either Microsoft Windows or Linux. The
controllers Analog-to-Digital Converter
(ADC) samples the temperature in the
cooler using a thermistor. Pulse Width
Modulation (PWM) outputs on the
controller board feed a driver board
that, in turn, drives a Thermoelectric
Cooler (TEC) and a pair of low voltage
heating elements. A single digital
output controls a fan that circulates
heated or cooled air within the cooler.
The whole system except for the
power supply is pictured in Figure 2.
Controller Board
Vellemans board comes either
fully assembled or as a kit. I chose the
kit to save money and found it was
easily constructed. The construction is
all through-hole, with socketed ICs and
no surface-mount parts. The board
comes solder-masked and silk-screened
for easy assembly. Screw connectors
are provided for the five digital inputs,
two analog inputs, two PWM outputs,
and eight digital outputs.
If youve never assembled one of
Vellemans kits before, you may be
pleasantly surprised at how easy it is to
follow their almost completely pictorial
instructions. The box is printed in
four languages, but the assembly
instructions just show the order in
which the parts should be installed. An
English PDF manual available from the
Velleman website explains the provided
software drivers [1].
The design of the board is simple
but efficient. Basically, all the input and
output functions the board provides
are handled using a PIC16C745
microcontroller that also handles the
USB interface. This microcontroller is
not user re-programmable, but it is
quite functional.
Among the nice touches provided
on the board are LEDs on all outputs,
Normally Open (NO) pushbuttons to
test the inputs, op-amps to buffer the
analog inputs and outputs, and a
counter function on two of the digital
inputs. Not bad for about $45.
Figure 3 shows the assembled
controller board.
Temperature Sensor
For the temperature sensor, I
chose a Negative Temperature
Coefficient (NTC) thermistor I found at
a local surplus electronics outlet. The
resistance of an NTC thermistor goes
down as temperature goes up in a
negatively accelerated, decreasing
function. If you can measure the
resistance accurately and if you know
the non-linear characteristic curve
relating resistance to temperature in
your particular thermistor, then you can
infer temperature.
Because our thermistors specific
characteristics were unknown, we
initially attempted a crude calibration
by comparing measured resistance
values against an available electronic
thermometer. Coupled with a software
table lookup, this yielded acceptable
performance in the initial classroom
prototype, but I wanted to achieve
FIGURE 2. The RoboCooler System. Modified Igloo cooler.
The TEC and its external fan are mounted on the side facing
the camera. The box containing the interface board is
partially hidden on the right side of the unit. Not visible
are the controller (unplugged) and the heater (inside).
FIGURE 3. Assembled K8055 USB Experiment Interface Board.
Wires at the right carry PWM and digital outputs to the cooler
interface assembly. Analog inputs are on the left, near
my thumb. The variable resistors near the alligator clip at the
top of the board are replaced with a fixed resistor as
described in the text.
Reed.qxd 1/10/2008 10:38 AM Page 32
somewhat better accu-
racy. If you dont have
access to a calibration
lab and you have to
calibrate your own
thermistor, you have a
couple of options. First,
you can purchase
thermistors with known
and guaranteed charac-
teristics. These will
come with data indicat-
ing the appropriate coefficients for
resistance to temperature conversion.
Even if youre dealing with an
unknown thermistor, determining the
Steinhart-Hart coefficients for your
particular thermistor empirically is a bit
tedious, but not difficult.
The technique has been detailed a
number of places, including Circuit
Cellar [2], so Ill spare you the detail, but
it can be done by measuring the resist-
ance at three or more known, widely
separated temperatures, and solving for
the appropriate coefficients using a
spreadsheet or a dedicated program.
Fortunately, one of my students
works in an applied metrology lab and
ran a complete calibration of our
unknown thermistor over the range of
45 to 90 degrees Fahrenheit. The most
interesting part of the calibration
results is the determination of the
four coefficients needed to compute
the Steinhart-Hart equation relating
thermistor resistance to temperature.
With these coefficients in hand,
computing the temperature directly
from the inferred resistance of the
thermistor was easy to realize as a C
function. Figure 4 shows a snippet of C
code to convert an ADC count into a
Fahrenheit temperature, using the
values from calibration. Figure 5 shows
the calibrated relationship between
resistance and temperature for the
sample thermistor.
Input Interfacing
Thermistor resistance, and hence
temperature, is inferred by inserting
the thermistor into a voltage divider
connected to the input of an ADC. The
voltage divider circuit uses a known
resistance as one half of the divider,
and applies a reference voltage to the
top end of the divider.
The ADC measures the voltage
at the divider junction, which is
proportional to the resistance of the
thermistor.
The thermistor can be placed in
either the upper or lower arm of the
divider, with a resulting alteration in
available range. Selection of the fixed
SERVO 02.2008 33
// Convert ADC count to Degrees F
// Based in part on: http://www.maxim-ic.com/appnotes.cfm/an_pk/1753
double countToTemp ( long count, double gain, double r1 )
{
// These are computed via calibration
static const double a = 0.00393401767 ;
static const double b = 0.000252561203 ;
static const double c = -0.000000252514329 ;
static const double d = -0.000000672358356 ;
double tempK = 0.0 ;
double tempF = 0.0 ;
double ohms = 0.0 ;
// Compute thermistor resistance (in megaohms) from voltage divider formula
ohms = ( ( ( ( 256.0 / ( ( double ) count ) ) * gain ) - 1.0 ) * r1 ) / 1000000.0 ;
// Apply Steinhart-Hart to get reciprocal of temp (Kelvin)
tempK = a + ( b * log ( ohms ) ) + ( c * pow ( log ( ohms ), 2 ) )
+ ( d * pow ( log ( ohms ), 3 ) ) ;
// Temp K
tempK = 1.0 / tempK ;
//Convert to degrees F
tempF = ( tempK - 273.16 ) * ( 9.0 / 5.0 ) + 32.0 ;
return ( tempF ) ;
}
FIGURE 4. C function
to scale the raw
thermistor count to
degrees F. Based on the
Steinhart-Hart equation
and the calibration
data obtained for our
thermistor, this
function computes the
temperature in degrees
Fahrenheit. Count is the
raw count from the ADC
(0 to 255), gain is equal
to 2.0 in this design, and
R1 is equal to 47K.
FIGURE 5. Relationship between
thermistor resistance and temperature.
This graph was generated by stepping
through a range of resistance values and
computing the temperature predicted
by the equation implemented in the
code snippet of Figure 4.
Reed.qxd 1/10/2008 1:15 PM Page 33
34 SERVO 02.2008
resistor for the divider is subject to a
number of trade-offs, including
thermistor self-heating that you can
find detailed in other sources. I placed
the thermistor on the high side of the
divider, and selected a fixed resistor
value of 47K. These are compromises
and almost certainly not optimal, but
they yield satisfactory results over
the range of temperature covered by
the system.
To make the circuit simpler and
a bit more stable against changes in
board temperature or mis-adjustment,
I re-wired the analog input circuit
of the Velleman board as shown in
Figure 6.
The modifications relative to the
original schematic consist of replacing
RV1 with a fixed 47K resistor, and
bypassing R3 by connecting the
thermistor directly across the SK2
jumper. Gain of the op-amp is set to
two by installing a 10K resistor as R8.
Since the 47K resistor was a 5% part, I
used its measured value of 47,500 in
the program calculations. Calculation
of temperature from the ADC count is
based on the techniques described in
reference [3].
Over a range of 45 to 90 degrees
Fahrenheit and coupled with the
eight-bit range of the Velleman ADC,
this combination of values yields an
average step of a little less than one
half degree per count, as shown in
Equation 1.
EQUATION 1
V
45
= R1/(R1+RT
45
)*Vref*Gain =
47500/(47500+232400)*5*2 =
1.7 volts
V
90
= 47500/(47500+70500)*5*2 =
4.03 volts
Counts per degree =
(90-45)/(((V
90
-V
45
)/Vref)*256) =
45/(2.33/5)*256 =
0.38 degrees/count
V
45
is the predicted voltage at the
input to the ADC at 45 degrees F. It
depends on the value of the voltage
divider formed by the fixed resistor and
the thermistor, and it is scaled relative
to the reference voltage applied to the
top of the divider and to the gain of
the op-amp feeding the ADC. The fixed
resistor is 47,500 ohms, and the
resistance of the thermistor at the
specified temperature is calculated
from the calibration data using the
Steinhart-Hart formula. V
90
is the
voltage present at the same point
when the thermistor is at 90 degrees F.
Given the other factors (such as
FIGURE 6. Revised AIN circuit for the
K8055. This revision replaces RV1 with
a fixed resistor and shows where the
thermistor is connected to the board.
R8 and R11 establish the op-amp gain
at 2.0.
FIGURE 7. Schematic of the TEC, heater,
and fan interface. Output connections
to pins on the K8055 are also shown.
Reed.qxd 1/10/2008 10:39 AM Page 34
self-heating affecting thermistor
calibration), the assumed accuracy
of this system is probably around
1 degree Fahrenheit over this
temperature range. While this isnt
high precision, it does compare
favorably with the precision of the
electro-mechanical thermostat this s
ystem replaces.
Output Interfacing
Figure 7 shows the schematic of
the interface board.
The PWM outputs on the Velleman
board are buffered with discrete tran-
sistors, arranged in an open-collector
configuration. Since the heaters draw a
little over one amp and the TEC draws
four to five amps, driving the coolers
output devices directly is not a good
option. Therefore, the controller
boards PWM outputs drive a couple of
hefty MOSFETs.
The PWM outputs that drive the
interface board from the Velleman
controller are open collector, so in this
simple design, the sense of the PWM is
reversed. In other words, when the
PWM pulses are off, the transistors on
the controller board are essentially
open-circuited, allowing the MOSFET
gates to be pulled up to the 12 volt rail
and turning on the load. Similarly,
when the PWM pulse goes high, the
transistors on the controller turn on,
pulling the MOSFET gate down and
shutting off the load.
This inversion of PWM duty cycle is
a bit odd, but is easily compensated for
by reversing the sense of the 0 to
100% values passed to the controller
board for each PWM channel.
There are several fans in the
design, but only one is uniquely under
program control. The fan inside the
cooler responsible for stirring the air to
better distribute the heating and
cooling load is controlled by a relay
operated off two paralleled digital
controller outputs. I used a fairly
current-hungry surplus relay because
that is what I had on hand, so I chose
to parallel a couple of the open-
collector outputs from the ULN2803
chip on the controller. The fan on the
outside of the TEC to help cool the
hot side is wired to the same PWM
output as the TEC itself and goes on
and off with the TEC.
Figure 8 shows the construction
details for the interface board.
The MOSFETs are rated for more
than enough current in this
application, but the one driving the
TEC does get fairly warm over time,
since with its on resistance of about
0.1 ohms, the five amp load of the TEC
dissipates up to 2.5 watts. In view of
this and a good bit of caution, I added
home-made heatsinks to the MOSFETs,
as well as a tiny fan to force air
through the housing covering the
interface board.
The heatsinks are leftover copper
electrical connectors salvaged from a
circuit breaker box. Each is screwed to
one edge of the MOSFETs TO-204 can
after the application of thermal
grease. Since it wasnt practical to
isolate these heatsinks from the
transistors cases, and since the cases
are electrically connected to the
MOSFET drains, I elected to tie the
heatsinks into the interface controller
with a bit of heat shrink tubing and
nylon strain-relief fasteners. This
explains the odd, sloping angle at
which the interface board is mounted.
Wiring on the cooler was done
point-to-point using screw terminal
blocks where possible. While these
are not the most reliable sort of
connection under vibration and strain,
they have the saving grace of making it
easy to test the individual
components with a multime-
ter and rewiring as necessary.
It is pretty obvious
when the TEC is running,
since its fan makes a fair
amount of noise. But since the
heaters dont dissipate enough
wattage to glow (a safety considera-
tion in the classroom), I added an LED
that illuminates the inside of the
cooler when the heat is turned on.
(The LEDs I used had integral current-
limiting resistors for 12V operation,
and these are not shown in the
schematic.) A window in the cooler
lets the curious peer inside without
disrupting the thermal operation.
I fused the primary 12 volt lines
with automotive mini-ATC style fuses.
Of course, like all fuses, these are
mostly a protection against fire in the
event of a wiring short, and cant
provide much protection for the
MOSFETs, given the relatively slow
speed at which a fuse blows,
compared to the speed at which a
MOSFET dies.
Software
My students jointly designed,
coded, and validated the basic
thermostat algorithm by simulating the
programs inputs and outputs with
console input and output under
Microsoft Visual C++. This enabled us
to get a prototype version of the soft-
ware completed while the hardware
was still under development.
The temperature control
algorithm is the simplest possible one:
Measure the temperature and if it
differs from the setpoint, turn the
cooling or the heating full on, as
appropriate. This is sometimes
SERVO 02.2008 35
FIGURE 8. RoboCooler
Interface Board Construction.
On the left of the enclosure
is the fan relay. The two
MOSFETs and their heatsinks
are immediately to the right of
the fan relay. The small fan on
the right-hand side of the box
just provides cooling insurance
of the heatsinks.
Reed.qxd 1/10/2008 10:39 AM Page 35
referred to as Bang-Bang control,
reputedly from the noise that
mechanical actuators driven by this
method would make as the system
changes state. This is exactly what the
electro-mechanical thermostat in most
household heating and cooling
systems does, so the RoboCooler is a
faithful if limited emulation of the
sort of heating and cooling problem
with which everyone is familiar.
Originally, conversion of the
raw ADC count to temperature
was done using a simple table lookup.
This is quick to execute, easy to code,
and provides a good example of
employing and searching arrays for
the students.
If the host here were a microcon-
troller with limited memory and
integer math, then retaining the
table lookup method for count to
temperature conversion would likely
be a good idea. But as we are blessed
here with a fast processor (the PC),
lots of memory, and full floating
point, I rewrote the temperature
conversion to apply the Steinhart-Hart
equation directly, as shown in the
listing. This is not terribly efficient, but
it strikes me as more aesthetically
pleasing and easier to understand
than the former table lookup.
Bang-Bang control could oscillate
rather rapidly about the setpoint, so
the program we developed adds simple
damping in the form of moving
average. Successive raw temperature
readings are shifted through a 10-
sample window in the form of an array
and averaged. Since the samples are
taken at one second intervals, the
average value used for comparison
with the setpoint is a 10-second
average. This is really just a very basic
low-pass filter, but it seems sufficient to
damp the system effectively.
The software also tries to steal a
little more efficiency at the end of
cooling or heating cycles by keeping
the fan running for a programmable
time after the TEC or the heaters are
cut off. This is rather like what some air
conditioning systems do to scavenge
cold or hot air from the ductwork.
The hardware would support
much more intelligent and powerful
control algorithms, of course. Since the
TEC and the heaters are driven with
PWM, they can be varied proportional-
ly and not just switched fully on or fully
off. This would allow more sophisticat-
ed software to implement say, a
Proportional Integral Derivative (PID)
algorithm for smoother and more
accurate temperature control. For the
benefit of my first-time programming
students, we stuck to a basic software
control approach even though
this doesnt take full advantage of the
hardware capability.
One of the less pleasant aspects of
the system software which I faced was
making the Velleman-supplied Visual
Basic (VB) Dynamic Link Library (DLL)
work with a C or C++ program.
Adopting a layered approach, I
was able to structure a reasonably
programmer-friendly function library
for the cooler system, with mnemonic
functions such as TurnOnHeat(),
TurnOffCooling() and ReadADCCount()
that could be incorporated to replace
the stub functions in the student-
developed simulation.
But the lower layer essentially
the bridge from this function library to
the VB DLL proved challenging and
lacked portability. While the code
available for this project on the SERVO
website (www.servomagazine.com)
will work under Microsoft Visual Studio
6.0, the code would not compile or link
FIGURE 9. A sample cooling run.
The top line shows the TEC turning
on and off, and the bottom line shows
the temperature in the cooler.
FIGURE 10. A sample heating run. The
top line shows the heating element
cycling, and the bottom line shows
the temperature in the cooler.
36 SERVO 02.2008
Reed.qxd 1/10/2008 10:40 AM Page 36
against the newest version of Visual
Studio. Apparently, this has to do
with the way that linkage between
dynamic libraries compiled in different
languages has been altered in later
compilers.
The solution was to port the code
to Linux. This turned out to be much
quicker and much less painful than my
efforts to that point in linking later
versions of the Microsoft tools to the
K8055 DLL.
The Linux drivers do not require
any DLL and instead supply a shared
library that understands how to talk
directly to the Velleman K8055
hardware via another library of USB
routines. I rewrote the upper layer of
our hardware interface in terms of the
new library functions and the student-
developed code compiled under Linux
with gcc, required few changes.
The Linux library faithfully
emulates each function from Vellemans
VB DLL and so you can still use the
documentation that came with the
board for a guide in understanding the
capabilities of the board [4].
One late but useful addition to
the software was the ability to log
information about cooler operation to a
disk file while the system is running. This
facilitated checking the operation of the
system when heating and cooling.
Results
With the luxury of a Linux-hosted
computer dedicated to running and
monitoring the cooler, I was able to
gather some data showing how the
system performs. These data were
written to a log file in comma-
separated values (CSV) format and
then read into and graphed from Excel
or OpenOffice.
Figure 9 shows a typical cooling
run for the RoboCooler. The top trace
shows the status of the TEC, with 1
being fully on and 0 being fully off. The
bottom trace shows the temperature
declining from a starting value of 88F
down to the desired temperature of
70F in about 30 minutes. Note that
over time the temperature bounces
around the setpoint and the cooling
turns on and off as the cooler reheats,
with a period of about four minutes.
This is the Bang-Bang algorithm in
action.
Figure 10 shows a heating test.
The interpretation of the graph is much
the same as for cooling. Note that a
change in temperature of about 18F
takes almost twice as long as for c
ooling. I attribute this to the low-
powered heating elements employed,
which total about 15 watts.
Modifications and
Extensions
As with any project, experience in
developing and testing the RoboCooler
leads to a number of suggestions for
extension and improvement.
First and foremost, the interface
logic should be improved so that
the hardware comes up in a safe
configuration. As readers may already
have surmised from the discussion of
the inversion of the PWM logic, when
initially plugged into the USB cable,
the cooler comes up with both
heating and cooling fully on. If we
were talking about say, a piece of
rotating machinery, this would be
objectively dangerous, while in a
cooler it is just annoying. The software
we developed does attempt to
SERVO 02.2008 37
Special thanks to the students in the
Spring 2007 section of Introduction To
Programming Concepts, and the help
and dedication of the computer lab
staff, without whom this project would
not have happened.
Thank Yous
Almost all the parts I used are surplus, but
similar items remain widely available in
both new and used versions.
1. The Velleman USB Experiment Interface
Board is available as either a kit or
pre-assembled from Apogee Kits at www.
apogeekits.com/usb_interface.htm
2. The TEC I used is actually a surplus item
originally designed for cooling what
else picnic coolers. But TECs, with and
without integral heatsinks are available
from sources such as:
http://skycraftsurplus.com/index.asp
www.goldmine-elec-products.com/prod
info.asp?number=G2201
www.allelectronics.com/cgi-bin/item/
PJT-6/search/40_MM_SQUARE_THERMO
ELECTRIC_COOLER_.html
www.alltronics.com/cgi-bin/category.
cgi?item=04U003
(You may have more success when
searching for these items by using their
scientific name: Peltier devices.)
3. There are a lot of options for heaters.
You could use a 12 volt halogen lamp, for
instance, which consumes up to 50W and
puts off a lot of heat. Another possibility
is a set of resistors in series/parallel to
yield the sort of dissipation you need.
One heater resource is:
4. Thermistors are available from:
www.allelectronics.com/cgi-bin/item/
THR-19/search/THERMISTOR,_30K_
PREPPED_.html
www.mouser.com/search/refine.aspx?
Ntt=thermistor
5. A wide range of MOSFET devices will
work in this and similar systems. One
source for MOSFETs in the same family as
the IRF143 is:
www.mouser.com
Mouser also stocks appropriate relays for
fan control.
Parts Sources
Reed.qxd 1/10/2008 10:47 AM Page 37
compensate for this condition, by
steering the outputs to a safe state
both when it initially loads and as the
program terminates. This, however is
far from foolproof and a hardware
solution would be much superior.
Second, it would be fun to try out
some more sophisticated control
algorithms on the system by
implementing some form of propor-
tional or even PID control in lieu of the
simple Bang-Bang approach used here.
Im planning to let my next C program-
ming class take a shot at implementing
that improvement.
Lastly, we could further improve
the software so that it can
automatically switch from heating to
cooling and back again as environmen-
tal conditions change. As currently
implemented, the software like your
living room thermostat is either in
heating or in cooling mode and
requires manual intervention to switch.
This was a fun and simple project
but my hope is that readers will find
at least a few tidbits that may be of use
in their own, and likely more
sophisticated designs. If you are an
educator, perhaps youll consider how
embedded appliance applications
might be used to increase your
students involvement, not just in
digital hardware courses, but in
computer science and programming
classes, as well. SV
Jerry Reed holds a Masters Degree in
Computer Science from Webster
University and is an Adjunct Professor
of Computer Programming and
Applications at Valencia Community
College in Orlando, FL. In his day job, he
works as Senior Systems Analyst for the
college, and spends much of his spare
time involved with amateur radio and
microcontrollers.
About the Author
[1] Vellemans documentation for the
K8055 board is at www.apogeekits.
com/PDF_Fi l es/Manual _K8055.
pdf and www.apogeekits.com/PDF_
Files/Usermanual_K8055_DLL.pdf
[2] Brian Millers Temperature
Calibration System in the May 2007
Circuit Cellar, does an excellent job of
explaining both TECs and thermistors.
[3] Maxims Application Note entitled
A Simple Thermistor Interface to an
ADC explains how to calculate the
sensitivity and range of thermistors
connected to ADCs via voltage
dividers; go to www.maxim-ic.com/
appnotes.cfm/an_pk/1753
[4] Linux software for the K8055,
including source code and build/
installation instructions is available
at http://libk8055.sourceforge.net/
References
38 SERVO 02.2008
You know it wants one. Go ahead, spoil your robot. Give it the brain and the brawn to take on other robots.
All modules shown actual size
TReX dual motor
controller $99.95
Orangutan LV-168 $59.95
eX
u
a
o
tr
ller:
T
R
d
l
m
tor
con
o
i
v
e
e
i
a
s
n
s
a
n
n
o
ativ
d
s
g
n
llow
i
t
n
t
s
it
in
tw
en
ra
io
on
ro
w
ch
g
be
e
c
t
l
(
C
)
a
l
g
o
t
g
n
d
R
or
n
a
o
v
l
a
e
a
y
h
o
u
s
s
r
a
S
-
3
r
as
n
c
r
n
o
e
i
l
(R
3
2
o
T
L
w
i
a
l
i
u
d
f ad
t
o
l
T
)
th
m
u
t
t
e o
di
i
n
a
f
at
r
s
e
u
e
.
O
ran
g
u
tan
-16
: f
l
-
L
V
8
u
l
fe
u
red
r
t
tr
l
r
at
obo
con
ol
e
erfect
or
lo
-
olt
g
p
w
v
a
e
r
s.
A
m
16
8
u
obot
T
eg
a
C
,
2
-ch
n
n
l
H
r
d
g
es
,
a
e
-b
i
2
L
D
,
bu
z
z
e
,
a
d
8
x
n

e

m
or
.
Find out more at www.pololu.com or by calling 1-877-7-POLOLU. 6000 S. Eastern Ave. 12D, Las Vegas, NV 89119
TReX Jr
$59.95
TReX Jr: all the great features of
the full TReX in a lower-power,
lower-cost package.
Reed.qxd 1/10/2008 4:35 PM Page 38
SERVO 02.2008 39
N
eedless to say, with the sheer,
concentrated weight of genius
crammed into that tiny, bamboo
paneled Tiki bar, answers to these
questions and many more besides were
soon flowing thick and fast. And from
these ramblings MechBash was born.
Mechwars Robot Combat in
Minneapolis (which has been holding
both combat tournaments and
entertainment events since 1999) and
BotBash LLC in Arizona (holding
combat tourneys since the misty days of
antiquity) decided to team up and put
together a traveling show to push the
idea of robot combat, and emphasize
the educational aspect of the activity.
This new entity the MechBash
Robot Combat Show would be
expensive to operate, but not as expen-
sive as an actual robot tournament,
and the entertainment and education-
al content could be more reliably
managed. It would be attractive
mostly to large clients like state fairs
and big trade-shows.
A format was hammered out and
after knocking on many doors, a
booking agency was finally willing to
take a flier on the idea.
However, after weeks of frantic
tap-dancing and heroic efforts on the
part of myself, Bob Pitzer, and our long-
suffering agent, we had one single,
magical, this-either-turns-the-tap-of-
success-or-buries-our-efforts forever book-
ing the 2007 South Carolina State
Fair in Columbia, SC. We all bore down
and got working to make it a success.
One of the problems that a robot
combat promoter faces is the sheer
weight and sophistication of the infra-
structure required to stay one step ahead
of the builders in the arms-race between
the robots and the containment system.
And in terms of doing live shows, there
is the added complication of time.
The cage has to go up and come
down in about a day, and the econom-
ics of a show (or tournament) simply
dont work out. Its all very well to build
a super massive mega-cage, but if it
takes five days to set up, the promoter
has to pay four more days of Staples
Center level rent, and consequently
loses $80,000 on the show. The result
is either a quick bankruptcy for the
promoter or if outside sources of
income can be found to delay the
inevitable a slow and ugly death.
The Mechwars cage rated to han-
dle even mega-weight combat robots
(390 lbs) goes up in about 10 hours
Newton launches a washer.
how can we get robot combat with all its associated cultural and educational
benefits back into the public eye? Another question that came up was how
can I get paid to blow things up, and yet stay out of jail?
by Jonathan Vandervelde
One night, a gaggle of robot builders were posing questions like
One night, a gaggle of robot builders were posing questions like
Vandervelde.qxd 1/10/2008 11:56 AM Page 39
40 SERVO 02.2008
with a crew of 12. Each individual com-
ponent is designed to be man-portable.
The floor plane acts as a tensile
membrane from barrier to barrier, and
the under-floor support structure is light-
ened by using the truss-and-plate model
rather that the far heavier four-legged
sub-unit concept of stage design.
It goes up relatively quickly, but in
the case of an east coast show
thousands of miles from its Minneapolis
home base, its still a bit dicey in terms of
time/effort. Luckily, MechBash also has
Bobs Mobile Automated Robot Combat
Arena (MARCA). While not rated for
full-on combat with robots above middle-
weight size, it has plenty of strength for
the kind of entertainment event we were
holding in SC, and it is super trick in both
appearance and functionality.
So, after six months of prep and
organization, at 3:00 pm on Monday,
October 9th, the MARCA finished up
an 1,800 mile journey by rolling onto
the SC Fairgrounds in Columbia. We
pulled a few pins, hit the hydraulics,
and by 9:00 that night, three guys had
largely assembled a fully functional 24
by 40 robot arena with integrated
halogen lighting and P.A., electronical-
ly controlled pneumatic drop-pits and
flame projectors, and two roof mount-
ed video screens (David Therriens con-
tribution to the effort) to show close-
ups of the action in the cage.
The MARCA is essentially
its own 40 trailer, with a
folded width of approx 8
and a folded height of 13.
The central section of ceiling
is fixed, with longitudinal
support trusses running from
front to back, while the sides
of the ceiling fold down to
just past the centerline of the
road wheels.
The cage floor also has a
fixed central section, with
sides that fold up inside the
whole package. At the event,
the ceiling wings are raised
to the horizontal position,
locked into place by support
members which reach the ground,
then a flexible curtain of steel-framed
lexan panels is clipped together and
suspended from the perimeter.
The floor then folds down hydrauli-
cally. The wall curtain hangs 18
inches outboard to the edge of the
combat floor and acts to soak up
energy from a flying piece of robot in
much the same way the netting behind
the goal posts soak up the energy from
a kicked football, dragging it to a halt.
We had numerous occasions to
verify this useful property of the
design, as quite a few 35 pound
microwave ovens, smaller pieces of
shrapnel, and even one complete robot
made it over the barrier into the wall,
only to be gathered in and dumped on
the ground harmlessly. Throwing stuff
against the glass is as it turns out
a big hit with an audience.
Eventually, we started to actually
stage our robot bits in such a way that a
high speed spinbot would throw shrap-
nel directly onto the front wall, causing
parents to jump six inches in their seats
and the kids to laugh themselves silly.
The MARCA is perfect for this purpose
as the battlefloor is relatively narrow
and therefore confines the action to a
spot fairly close to the bleachers.
With the MARCA in place, the sec-
ond wave could come rolling in and by
8:00 am, the Minneapolis MechBash
crew had backed up another truck and
offloaded 12 big robots with the entire
MechBash educational robot program.
This includes a horde of individual R/C
and chassis components for kids to
assemble, and an ant/beetle cage with
functional hazards. Then came the
piece de resistance ... Bob unveiled the
robot co-host and mascot that he and
partner Chris Harriman had been
working on back at his Arizona lair:
Gizmotronic, a 5 8 humanoid (more
or less) with articulated arms, head,
and waist, a video screen face that
echoes a miked voice, and enough
sophisticated safety code to allow us to
drive it around and interact directly
with kids in the audience.
A robot display area was set up
along with a hands-on small-bot area
and merchandise booth. The MechBash
show was almost ready for business.
The MechBash Traveling Robot Combat Show
Big Bots on display
between shows.
Brave scout sits on Santa
Gizmos lap.
At 3:00 on a Wednesday after-
noon, the very first MechBash robot
combat show opened to a full house
and went off without a hitch. From
that moment on, we did four big
shows and countless hours of hands-on
mini robot classes every day for 11
days. Over the course of 47 big shows,
there were only four that werent
standing-room only, and we were
approached by numerous grade school
and high school instructors (and
students!) who were interested in the
robots from an educational point of
view, or were just interested period.
We were also approached by
dozens if not hundreds of adults
who wanted information on compet-
ing and wanted to know where the
nearest battle was. Luckily, MechBash
had skilled volunteer help from Chuck
Butler of Carolina Combat Robots, who
had brought a knee-high stack of fliers
for their next event, so we were able to
point people to a tournament within
driving distance.
By Friday of that first week, we
were into the second printing of
Chucks fliers, and the third printing of
our own getting started handouts.
Robot trading cards were going like
popcorn!
Meanwhile, with the highly skilled
help of Robert Woodhead (of
Animeigo) and family, we started to
record some show video.
A MechBash show actually starts
an hour before the curtain rises. The
projection screens light up with video of
robots battling, and people start to
wander over and sit down in the stands
to watch. The music starts to slowly rise.
Activity at the hands-on mini-robot cage
starts to grow as more people begin to
congregate, and Dr. Tim shows them
how to assemble an ant-bot using
hacked servos (as drive motors) and a
Vex radio or older Futaba rig.
Children poke at Gizmo and follow
him as he starts to roll around the site,
waving his arms and chattering. Pretty
soon, we have to station a man to help
with the ant-weight arena and another
man to take pictures of the kids with
Gizmo. We also have to station a robot
wrangler at the big bot area
to answer questions. Half an
hour from show time, the
robot wranglers push the
robots chosen for the next
show into an on-deck que
in front of the cage.
These robots have been
designed with three rules in
mind: They have to look
cool; they have to stand up
to roughly 50 consecutive
matches with minimal
repair; and they have to be
able to do something exciting.
(Also, it helps if theyre big.)
Many people who have
seen robot combat on TV or
on video dont realize how
large a 340 pound bot actually
is. The sheer size of the bigger machines
never fails to excite the audience.
Shock-mounting batteries and
electronics, using belts on the motors
rather than chain or shaft drive, heavy
wiring with plenty of breakers, these
are all good rules of thumb for building
either a real combat bot or a show bot.
With a show bot however, looks
are critically important. Lots of
empty air inside the hull is usually
a good thing (more size without
much weight), and it must be able to
do something cool.
At curtain-time. the music sudden-
ly rises, the sound man triggers a.wav
file from the movie Aliens (Warning,
the self destruct mechanism has been
activated ...), and the M.C. pops out
from the curtain and exhorts the audi-
ence to cheer for robotic destruction.
Each of the MechBash robots is
introduced, some mechanical or
electronic design considerations are
discussed (as a way to bridge the set
up period), then the first machine
SERVO 02.2008 41
Dr Tim lets kids assemble a vex-bot
while the Hockey-bots cool off.
The ladies love Gizmo.
Vandervelde.qxd 1/10/2008 4:21 PM Page 41
42 SERVO 02.2008
destroys a target, races up a ramp and
jumps something (and destroys it), or
runs a race (and destroys something).
Video screens above the cage show
close-ups of the action to the 30 or 40
people who couldt get seats and are
standing at the entry end of the site
trying to see.
As the show goes on, people
are pulled from the audience and get
to operate the flamethrower while a
wrangler drives the robot, or they get
to drive a weasel-car, and try to escape
destruction, or they actually get to pilot
a big bot and face off against another
audience pilot.
If were doing it right, the cheering
gets louder and the audience gets
rowdier with each individual bit, until
the big, cataclysmic finale brings down
the house, and people come streaming
out of the stands to ask questions, to
touch the robots, or just to stand
around and grin.
The tricks to a robot combat show
are to keep the audience engaged, to
develop a good rhythm, and most
importantly to get an exciting bang
out of your equipment. Its easy to sim-
ply break a microwave with a robot; the
trick is to smash it in an exciting way.
High speed spinners are a good
bet, but reliability is always an issue,
and building a spinner that can do this
particular job is not as easy as you
might think.
Our closer for these shows was a
MechBash housebot named Newton.
Newton was built by Tim Wolter and
his son Karl, with some initial sugges-
tions by yours-truly. Newton runs a
single 3 hp Magmotor on a belt drive,
pushing 50 pounds of steel rotor, with
a diameter of close to 4. The rotor is
massive, but has a relatively thin, sharp
section at the business end. Full spin-up
time was about 15 seconds, but in the
context of a show bot, this is not an
issue. In fact, the robot was intention-
ally designed with a long spin-up time
and variable-length rotor arms in order
to demonstrate how potential energy
and momentum are affected as these
conditions vary.
We typically used Newton in the
big finale to rip a dishwasher in half
another audience favorite. As an inter-
esting comparison, we also brought
along an actual mega-weight combat
robot Stump-grinder from Team
FUBAR as our Newton back-up.
Stumpy has dual mag motors,
similar gearing, and 65 pounds of flail,
so on paper its performance should
have left Newton in the dust. In terms
of entertainment however, we
discovered this problem: On a large tar-
get like a washer, Stump-grinder would
simply cut a big slot through the base.
The object hardly moved and shrapnel
was minimal. With a smaller target,
Stumpy would loft it across the cage,
but create relatively little shrapnel due
to the fat cross-section of the flail.
Functionally, the robot is great for
combat, but we struggled with it in a
show context. Eventually, we (acciden-
tally) found Stump-grinders mtier by
slipping up and driving through the
intended target, completely obliterat-
ing the safety backstop of 15 cinder
blocks, and making an unholy racket
and massive clouds of dust and debris.
This was an exciting bit which the
crowd really enjoyed, but nearly caused
a rebellion by the robot wranglers, who
had to sweep up the mess.
Our early shows also incorporated
the holy grail of robot-combat
entertainment combat robots that
actually battle during the show. Our
U-fight-ems, version 2.0, lasted about
three days before robot-A got toasted.
Robot-B was subsequently driven over
the barrier and onto the floor by an
eight year old girl. The MARCA did a
beautiful job of containing it at the
wall, and the machine survived, but at
that point we decided to pull them out
of the line-up for re-tooling.
The issue that weve been
struggling with, other than keeping
them in the cage is giving them
active weapons that continue to
hammer away in a constant fashion,
and yet dont actually toast the robots.
You want to be able to smash a target
with the robot, proving its mightiness,
then turn around and let two of
them actually battle, serene in the
confidence that they will both survive.
At the same time, something has
to appear to happen to one of them, or
there is no closure to the match. Tricky,
but not impossible. We are currently
close to success. Version 3.0 will
probably last four or five days of
combat, but when we get to version
4.0, look out world!
The MechBash Robot Combat
Show was the hit of the 2007 South
Carolina State fair. We had just under
10,000 spectators in the bleachers over
the course of the show, and several
times that when you include hands-on
participants and video watchers.
Response from education professionals
was through the roof, and we burned
through multiple printings of fliers
promoting the nearest RFL event (the
North Carolina machine battles).
MechBash is already confirmed for
another major state fair event next
summer, in talks with two more, and
up for any number of other potential
gigs in the fullness of time. We hope
that interested builders, designers, and
conceivers will get onboard with us as
we try to push forward with this model
and continue to promote robot com-
bat, engineering, and inventing. SV
Newton smashes a Xerox machine.
SERVO 02.2008 43
W
hat do robots have in
common with recycled
clothes, root beer, and
medieval siege weapons? Why are
the creators of these wonderfully
interesting things suddenly more visible
than they were in the past? What is
Hyperbolic Origami? These are a few of
the questions I pondered as I drove
from Dallas to Austin, TX for the 2007
Austin Maker Faire.
I had just returned from a road trip
to Marfa, TX for the annual Chinati
Open House art festival. It struck me as
I took in the art and talked to the
artists in Marfa that the robot builders
I interact with every day arent really
that different from artists. Theyre both
creative, both a little eccentric, but I
couldnt quite put my finger on the
common thread that connected the
two. There was some growing
convergence or similarity that I hadnt
previously noticed. The Austin Maker
Faire promised a large supply of new
sample data for me to observe, from
both the art and robotics communities
that I was already familiar with, as well
as the craft community, with which Id
had little previous experience.
The weather in Texas was perfect
for a long road trip and there were two
beautiful sunny days for the event
which was held at the Travis County
Expo Center. Even though I arrived
the evening before Maker Faire, I got
into the spirit of things immediately.
My hotel room was strangely dark,
which I soon discovered was due to the
four lamps in the room having 25 and
40 watt bulbs. Why complain when
you can do it yourself? At a Home
Depot down the road, I bought
four 100 watt bulbs and upgraded my
hotel room.
Later that evening, I learned I
wasnt the only one with a do-it-your-
self attitude at my hotel. I met a cat
who had learned that the motion
sensor on the hotels lobby doors
was sensitive enough to activate for a
MAKER
F a i r e
by R. Steven Rainwater
Rainwater.qxd 1/10/2008 10:55 AM Page 43
small furry mammal. The cat took
advantage of this anytime it got
hungry, walking right into the lobby as
if it owned the place and caging treats
off the guests.
The next morning, I set out for
Maker Faire with camera and notepad,
not really knowing what to expect. My
first impression was of a smaller scale
event than I had imagined, but my
disappointment soon turned to
amazement as I realized how much
was packed into the fairgrounds.
Maker Faire must be judged on a
different scale than trade shows and
technical conferences. Ive covered
seemingly larger events in a day. Two
days was insufficient to see even a
fraction of Maker Faire. One big
difference here is that almost
everything at Maker Faire requires your
participation or interaction. Dont
expect to just walk around and gawk
at things like you might at a trade
show. At Maker Faire, the line between
spectator and exhibitor is blurred.
Want to ride the Cyclecide
Carousel? You can bet youll have to
pedal. See an electronics kit you want
to buy? Youll be provided with tools,
test equipment, and space to assemble
it. Standing too close to the girl who
builds electronic synth gear out of
childrens toys and surplus medical
equipment? Careful or shell put you to
work hacking toys. Admiring the art
cars? Youll be invited to join in on
construction of a new one. Donate a
bag of old clothes to the Swap-O-Rama
and youll soon find yourself cutting,
sewing, and silk-screening a new outfit,
with expert help if needed. Maker Faire
is very much a DIY event in every sense.
There are a few exceptions. Youll
have to keep your distance from noisy
machines belching flame and sparks in
the ring of fire area. And while you
may be asked to help turn the cranks to
hoist the 4,000 lb. safe into the air
during the execution of the life-size
Mousetrap game, youll have to keep
your distance when it plummets to the
ground with an impact that can be felt
a hundred yards away.
Everywhere you look at Maker
Faire, youll meet interesting people
willing to stop and explain how their
creation works, or how they made it,
or why they made it, or who did their
tattoos, or answer any other questions
you have. Very quickly, I learned that
Maker Faire is as much about the
creators as it is about their creations.
As I talked to the creators or makers,
as we were encouraged to call them
it came through clearly to me that
most were motivated by a desire to
create something unique rather than
simply buying a mass produced
equivalent. Interestingly, the few
companies that showed up with mass
produced goods such as Ugobe with
their Pleo robot understood this
motivation and showed off ways in
which their products could be hacked
and customized.
Speaking of the Pleo, youre
probably wondering when Im going to
get around to telling you about all the
cool robots that were at Maker Faire.
The Pleo is a good place to start. There
was not just one Pleo but a dozen on
display. Some had their rubber skin
stripped off to reveal the mechanics to
curious robot builders. John Sosoka
was there answering technical
questions. There was even a one-of-a-
kind Pleo mod with snowy white fur. (It
looked more like a llama than a
dinosaur.) The Pleo looks like a toy with
a lot of potential for robot hackers.
Several other robotic pets could be
spotted throughout the fairground: a
robot horse, robot cat, a monkey, even
a parrot. Like the Pleo, these were
robots disguised as stuffed animals
but, unlike the Pleo, they were less
sophisticated. In every case, the exhibit
included both a standard model and
a model with the skin removed to
make it easy to assess the hackability of
the toy.
Most of the robots at Maker Faire
were one-of-a-kind creations built by
individuals. A nice selection of student-
built robots could be found at the BEST
(Boosting Engineering, Science &
Technology) booth, which included a
full BEST contest course. Curious
spectators watched robots made from
wood bases compete against robots
held together by PVC pipe, moving
bright orange cubes around the
course. More advanced robots built by
UT Austin students were roving around
the main arena.
MAKER Faire
44 SERVO 02.2008
Robot monkey, half with fur
and half without.
Austin Robot Group
Babbling Head, an
animatronic art
head that sings
sea shanties.
Austin Robot
Group sculpture.
Robots frequently turned up in
unexpected places. Exploring the show
barn, I noticed a modified iRobot
Create robot platform painting a
rectangular canvas. The art robot
was built and programmed by Kris
Degraeve. She and her sister, Carly,
had driven to Austin from Iowa to
show off their robot. Theirs was
just one example of a robot built by
female roboticists, in case anyone is
still under the impression that robot
building is a hobby limited to
males. Puiyee Hung, another female
robot hobbyist, was one of many
members of the Robot Group of
Austin demonstrating and explaining
their homebrew robots.
Austin is The Robot Groups home-
town and they were out in force with
all sorts of robots: autonomous robots,
remote-controlled robots, robots that
made art, even art that looked like
robots. They also brought along
several of their famous Thereping
musical instruments, which use an ultra-
sonic sensor to simulate the characteris-
tics of a Theremin. (See the April 06
issue of Nuts & Volts for the full
project.) Real Theremins could be found
elsewhere at Maker Faire, which includ-
ed a surprising number of individually
made musical instruments of all sorts, as
well as three stages with live music
throughout both days of the event.
If robots built from food are more
your style, the folks from Evil Mad
Scientist Laboratories could be found in
the show barn. They had created a vari-
ety of pumpkin robots including a very
detailed, motorized Dalek that moved
around under remote control while
menacing spectators wielded weapons
improvised from kitchen utensils. They
also created a Cylon pumpkin with a
traditional red vision sensor moving
back and forth across its face.
Nearby, other pumpkins met a
SERVO 02.2008 45
MAKER Faire
Austin Robot Group's spooky
Halloween robots.
iRobot CREATE Art robot
built by Kris and Carly.
Alix plays a Thereping
(www.thereping.com).
BEST robots compete to
move orange blocks.
Austin Robot Group SanDraw
sand drawing machine.
quicker but less artistic fate; they were
used as projectiles in the medieval
siege weapon contest. Watermelons
too were hurtling through the air
behind the show barn, propelled out
of large wooden catapults and
trebuchets. After each shot, children
ran to the target area to call out the
distance achieved.
Having learned a great deal about
the non-traditional uses of gourds, I set
out for the Maker Store, passing
through the food makers area along
the way. I stopped to sample Maine
Roots root beer. Maine Root is a small
Austin company that makes root beer
the way God intended, with actual
cane sugar instead of the foul-tasting
high-fructose corn syrup used in
mass-produced root beers like IBC and
Barqs. I consider myself a connoisseur
of root beers and put Austins Maine
Root root beer in the same class with
New Jerseys Boylan and Louisianas
Abita, two of my favorites. I also tried
Maine Roots latest creation, a pink
soft drink made from juice of the Texas
prickly pear cactus.
Finally arriving at the Maker Store,
which also housed the Make and Craft
Labs, I ran across more robots; this
time, a table full of small Solarbotics
robots including Mousebots, Turbots,
and SolarSpeeders. In addition to
playing with working robots, kits could
be bought and assembled on the spot,
using provided tools. People of all
ages were building robots at the
workbenches, many for the first time.
While watching the busy robot
builders, I noticed one wearing a
Chinati 2007 T-shirt. I wasnt the only
person whod gone from admiring art
in Marfa, TX to building robots in
46 SERVO 02.2008
MAKER Faire
Reconfigurable sculpture based on
stop-motion armatures by Matt Norris.
A student-built robot in the
BEST competition.
DO NOT put your feet
down mid-spin or they
will snap off like twigs.
World's Fastest Bar-stool.
Austin Robot Group D2 Robot.
Austin, TX. This brought me full circle
to the questions Id arrived with. Why
are the makers of art, music, robots,
even food and clothes, coming
together in larger numbers and with
increasing frequency? With this thought
in mind, I noticed on a nearby white-
board that Wendy Tremayne, founder
of Swap-O-Rama, was giving a talk in
the main arena titled The Maker as
Revolutionary. Intrigued, I headed that
way to see if she had anything to say
that might tie up some of the loose
threads between artist, robot builders,
and the whole DIY movement.
Wendy made the point that big
corporations, in their continual effort
to maximize profit, have pushed us all
into becoming nothing but consumers
of mass produced goods. From the
clothes we wear and the food we eat,
to the music we listen too, its all mass
produced. One of the side effects is
that we lose our freedom to be unique.
Instead, we simply choose from
options provided by corporate market-
ing experts. Everything they sell us is
designed to go out of fashion as
quickly as possible, so well have to
discard things and replace them
with newer things were told are more
fashionable, further boosting profit.
Most people who stop to think
about it agree this excessive consump-
tion is a bad thing. Usually, its the
waste and pollution generated by the
endless consumption and discarding
of products that people notice. The
other costs are less obvious: loss of
community, creativity, and personal
freedom. This point brought me back
to something familiar. There is one
community that started doing some-
thing about the loss of freedom more
than 20 years ago. Richard Stallman
started the free software movement
with the idea of replacing proprietary
software with software that restores
the users freedoms to examine,
modify, and share software. Sharing,
which was considered a virtue by
previous generations, is anathema
to corporations trying to maximize
consumption. Today, were bombarded
with advertising that tries to convince
us that sharing is bad. In many cases,
it has actually become illegal! Free
software (or open source software as
some call it today) gets around these
problems, making sharing legal and
virtuous again.
Wendy suggested that other
communities have noticed the success
of the free software movement and are
beginning to look for similar ways of
combating the consumption that
dominates our society. Her own idea
was to create the Swap-O-Rama as a
way of fighting back against the
fashion industry. That each of us has
the freedom and capability to make
what we want, the way we want,
whether its a musical instrument, a
robot, art, or our clothes is a powerful
idea. Its an idea thats spreading into
every imaginable field. Despite being
a member of the free software
community for many years, I hadnt
connected its ideals with what was
happening in these other communities.
What will the proponents of
consumption think of these people,
these makers? Theyre a threat to the
status quo, of course. Theyre a threat
to corporate profits. Theyre revolution-
airies! You probably never realized that
building robots had such far-reaching
ramifications.
Maker Faire was coming to a close
and there were so many things I hadnt
had time to do or see. I still didnt
know what Hyperbolic Origami looked
like. As I left the fairgrounds that
afternoon, I passed several people
wearing T-shirts with one of the Make
slogans. I cant think of a better
message to leave my fellow robot
builders with: Void your warranty,
violate a user agreement, fry a circuit,
blow a fuse, poke an eye out ... SV
If you spent any time at the Austin
Makers Faire, you probably noticed two
cute girls watching over an artistic robot
as it busily painted a canvas. That would
be Kris and Carly DeGraeve. Between
the two of them, Kris and Carly seem to
be solely responsible for 90% of the
artistic output of Iowa. They create jew-
elery, clothes, and fashion accessories.
Visit any of their several websites and
youll also find paintings, photography,
art books, T-shirts, and too many other
creative things to list. Until recently,
robots were about the only item missing
from their repretoire.
Kris had been interested in art mak-
ing robots for a while when she noticed
a contest at the instructables.com
website. As part of the contest, iRobot
offered several scholarship packages
that included an iRobot Create robot
with command module. She submitted a
proposal and won a scholarship kit.
In very little time, Kris was writing
programs, modifying the iRobot Create
with paint handling hardware, and
teaching it to create art using a variety of
brushes, rollers, and sponges. She was
particularly fascinated by human-robot
collaboration. Rather than using the
robot as a tool to create art, or allowing
the robot to create art purely
autonomously, she settled on a collabo-
rative method that involves creative
input from both human and robot.
With their newly created art robot,
Kris and Carly drove all the way from
Cedar Falls, Iowa to Austin, Texas for
Makers Faire where they found an appre-
ciative audience. Their exhibit won a
prize from Craft magazine and nearly
everyone who saw it stopped to admire
the robot, take photos, and talk to the
girls about their work.
Since Makers Faire,
Carly has been pattern-
ing and sewing hood-
ies. Kris is taking a break
from robots to do some
painting on her own
but shes already think-
ing about new robot
projects and improving
her new programming
skills. To find out the
latest on what Kris
and Carly are up to,
visit their website at
www.krisandcarly.com.
Crafty Girls
MAKER Faire
SERVO 02.2008 47
Kris and Carly with
their art robot.
48 SERVO 02.2008
T
hat figures as for the past six
months, Ive been working on
an embedded project that is
dependent upon the motion provided
by the shaft of a stepper motor. So, it
would be an understatement to say
that Ive been heads down in the
design and implementation phases of
deploying the embedded motor control
application Ive been sleeping with for
the past six months.
Along the way, Ive had the oppor-
tunity to build up and test various step-
per motor control circuits. Im on the
verge of building up yet another step-
per motor controller that is based on
the STMicroelectronics L6208, which is
a single-IC DMOS driver for bipolar
stepper motors. The L6208 can be had
in a trio of packages: PowerDIP24,
PowerSO36 and SO24. Ive selected
the PowerSO36 package and if you
would like to see how this project
comes out, keep reading, my friend.
The STMicroelectronics
L6208
A close friend of mine owns and
operates a machine shop, which contains
a multitude of large lathes and milling
machines. I can recall walking into his
shop one afternoon to find him pouring
over a very large piece of electrical gear.
Between his cussing and swearing, I
concluded that the large collection of
very large electronic components he was
working on was part of one of his mills.
When I asked him just what the problem
was and what he was working on, he
explained to me that this particular mills
stepper motor had seemed to fail.
Obviously, the easy thing to do in
this situation is replace the stepper
motor before looking elsewhere for the
problem. So, he had already swapped
out the stepper motor with no joy by
the time I had gotten there. Mind you,
the stepper motor in the failing mill
was quite large. With the stepper
motor out of the failure loop, the next
logical place to check was the mills
power supply. According to my buddy,
every voltage was present and account-
ed for. Thus, my machine-head friend
was troubleshooting the stepper motor
driver electronics when I arrived.
To this day, I envy his courage and
adventurous spirit as I would have
NEVER removed that massive collection
of high-voltage, high-wattage transform-
ers, resistors, and capacitors. Although
most of the sensitive control electronics
were stuffed onto printed circuit boards,
the bulk of the stepper motor driver
consisted of large high-voltage/high-
amperage electronic components.
If youre into machine shops, you
know that lots of the mills in smaller
shops were obtained second-hand
by Fred Eady
Building a
STEPPER MOTOR
CONTROLLER Part 1
PHOTO 1. This over and under shot gives you a total physical view of the L6208PD.
The L6208PDs heatsink tab runs the length of the bipolar stepper motor driver IC.
Im always thinking of the reader (thats you) and with this type of package, Id better
make the project easy to build and reliable as lifting a toasted L6208PD from its printed
circuit board heatsink pad wont be any fun at all.
Eady.qxd 1/10/2008 11:42 AM Page 48
from large machine shops and are usu-
ally of older vintage and dont contain
up-to-date technology. So, most folks
buy the old mills and upgrade them
with modern electronics. After finding
out that replacing the mills original
antiquated stepper motor controller
would cost as much as the mill did, my
friend retrofitted the mills electronics
with a couple of stepper motor drivers
the size of a couple packs of cigarettes.
Its not my intent to design and
implement a mill stepper motor driver.
However, I do want to get the most
bang for my stepper motor buck from
the smallest possible package. Free
space on efficiently designed robotic
equipment is equivalent to free SRAM
on small microcontrollers they are
both usually hard to come by.
The size versus power factor was the
key to my selection of the L6208 in the
PowerSO36 package (which is ordered as
part number L6208PD.) The raw
L6208PD we will be building our stepper
motor controller project around is shown
up close and personal in Photo 1.
As I stated earlier, the L6208PD is a
DMOS bipolar stepper motor driver that
incorporates everything necessary to
interface a bipolar stepper motors coils
to a logical motor controller. In this
case, the logical motor controller will be
a Microchip PIC. Well make a decision
on which PIC well use after we work
out the details of the L6208PD stepper
motor design. An STMicroelectronics
microcontroller would be a good choice
here, as well, but I dont have access to
any of the STMicroelectronics microcon-
troller development tools. I do have an
extensive set of Microchip development
tools. So a PIC it is.
Obviously, not everyone that picks
up a SERVO magazine holds a
doctorate in mechanical or electrical
engineering. However, Im sure some
of you out there are pretty tricky with
electromechanical things. Regardless
of your mechanical or electrical
prowess, Im going to explain in
basic terms what really needs to be
explained in order for you to be
successful with your personal L6208PD
design. So, during our stepper motor
project discussion, I will not assume
that you are a stepper motor expert. In
my dealings with the customers I
design gadgets for, I am often
described as an expert and I quickly
remind the purveyor of the compliment
that an expert to me is simply a drip
under pressure. With that thought,
lets talk about bipolar stepper motors.
A bipolar stepper motor interfaces
to the L6208PD with four leads. The
four leads of the stepper motor are
designated as A, A- (A bar), B, and B- (B
bar). In most stepper motor data
sheets, the bar designations have a
dash above the letter. For instance, A
bar on the stepper motor datasheet will
be denoted as A. The stepper motor
phase output nomenclature for the
L6208PD is defined on the L6208PD
datasheet as 1A, 1B, 2A, and 2B.
To make the stepper motor rotor
spin under our control, we must match
up the stepper motor datasheet phase
designations to the L6208PD datasheet
phase definitions. Dont worry about
that right now. Well figure out the
stepper motor phasing after we finalize
the stepper motor controller design.
However, we can make a decision on
the stepper motor we will use. Ive
settled on the Minebea 29SM-K7
stepper motor you see in Photo 2, which
is configured for bipolar operation.
When it comes to understanding
how a bipolar stepper motor works,
one should envision a pair of simple
coils positioned 90 relative to each
other that are using their magnetic
fields to influence a permanent magnet
rotor. To bring that last statement to
life, Ive sacrificed a stepper motor for
your inspection in Photo 3.
As you can see, the eviscerated
stepper motor in Photo 3 has more
than a single pair of coils and a simple
permanent magnet rotor. A close exam-
ination of the quartet of rotor magnets
revealed that each rotor magnet
consists of 50 notches and each rotor
magnet was rotationally offset from its
neighbor magnet by one notch.
Im sure theres a mathematical
link to all of the stepper motors physi-
cal mechanical weavings, but its really
not important to our hardware design
as the L6208PD is designed to drive the
stepper motor with little effort from us.
SERVO 02.2008 49
PHOTO 3. Yep. There are more than two coils here. This motor is wired to provide a single
step rotation of 1.8 per step versus the 45 per step of our logical stepper motor model.
PHOTO 2. This is a hefty stepper motor. It
is designed for high torque at relatively
low rotational speeds.
Eady.qxd 1/10/2008 11:43 AM Page 49
50 SERVO 02.2008
From what I can tell by visually tracing
the wiring between the stepper motors
eight coils, each of the stepper motor
coils is electrically connected to form
the logical two-coil-and-a-rotor stepper
motor model presented in Figure 1.
The rotor in Figure 1 is aligned at a
45 axis point because equal current is
being applied to each of the stepper
motor coils. If we reverse the current in
the correct coil, the rotor will rotate 90
with each current reversal operation.
This 90 movement of the rotor in our
logical example is typical of full step
mode. You can see the logical current
reversal patterns that influence the
stepper motors rotor in Figure 2. The
step angles can be halved by manipu-
lating the presence of and direction
of coil current during a step period.
Our logical stepper motor is
super simple. However, in the real
world a typical stepper motor will
contain multiple poles (like Photo 3),
which will increase the number of
steps that are required to complete
one full revolution of the stepper
motor rotor. The stepper motor we
will have selected to drive with our
L6208PD-based stepper motor
controller moves in 1.8 steps,
which equates to 200 steps per
full revolution.
The L6208PD is designed to
drive a two-phase bipolar motor like
the one weve chosen in Photo 2. As
you can see in Photo 3, our stepper
motor isnt the simple two-coil motor
described visually in Figure 1. However,
we hook up the stepper motors four
wires to the pair of L6208PD H-bridges
as if it were that simple.
A long time ago, I actually wrote
some PIC code and built some ULN2003-
based H-bridge hardware to directly drive
a pair of small stepper motors on a crude
little hit-something-and-
backup robot I was play-
ing around with. We
have been spared the
labor of writing step
code as the technology
contained within the
L6208PD automatically
generates the correct
phase patterns that
result in the correct
stepping sequences for
the stepper motor that
is attached to the
L6208PDs H-bridge out-
put pins. With the flick
of a bit, the L6208PD
can drive a suitable
stepper motor clockwise
or counterclockwise
in full step and half
step modes.
Walking Around the
L6208PD
Ever sit in an airport concourse
and watch the pilots perform the ritual
preflight walk-around? The pilot is
looking for the obvious bad things
that may cause problems with the jet in
flight. When we do a walk-around,
were looking at what needs to be
done to the peripheral pins of an IC to
bring it to life in our design. Lets per-
form a walk-around on the L6208PD.
Usually, the more pins an IC has the
more complex it is to use. That axiom
doesnt hold true for the L6208PD. Of
the L6208PDs 36 pins, 12 are not
connected, four are ground pins, and
two pins let power flow into the
L6208PD. That leaves us only 18 pins to
be concerned with from a hang-some-
thing-off-it standpoint. We can further
reduce the L6208PDs design complexi-
ty by designating the L6208PDs RESET,
CLOCK, CW/CCW, HALF/FULL, and
CONTROL pins as direct logic inputs that
need nothing but to be directly connect-
ed to a host microcontrollers I/O pins.
A logical low applied to the
L6208PD RESET pin places the
L6208PDs phase sequence generator
state machine at what the L6208PD
datasheet calls State 1. State 1 is
defined as a condition in which the
CLOCK input is logically low and the
currents flowing through the stepper
motors A and B coils are equal. State 1
is also referred to as HOME State.
The L6208PD CLOCK input is fed
into the phase sequence generator
state machine and a step is generated
on each rising edge of the incoming
step clock signal. The maximum clock
frequency we can apply to the
L6208PD CLOCK pin is 100 kHz.
Its rather obvious what the
CW/CCW pin has to do with the rotation
of a stepper motors rotor. A logical high
level applied to the L6208PDs CW/CCW
pin results in clockwise rotor rotation.
Conversely, a logical low at the CW/CCW
pin causes the stepper motor rotor to
spin in a counterclockwise direction.
The L6208PDs HALF/FULL pin
function is also just as obvious. The
L6208PD step mode is determined by
the logic level applied to the HALF/FULL
FIGURE 1. Programmers that arent hardware
gurus need abstract hardware models of the
target microcontroller or microprocessor to
fashion their firmware. Programmers that
arent mechanical engineers will need this
logical figure to grasp the basic concept of
how a stepper motor works.
FIGURE 2. This is highly simplified. However, what you
should note is the logical progression of current reversal
operations and how they influence the rotors position.
Eady.qxd 1/10/2008 11:43 AM Page 50
pin. A logical high level at the
HALF/FULL pin puts the L6208PD into
HALF STEP mode. FULL STEP mode is
entered when the logic level on the
HALF/FULL pin is low. The logic levels
for CW/CCW and HALF/FULL
operation are easy to remember
as the functions are selected as
HIGH/LOW. That is, CW and
HALF are active when a logical
high is applied to their respective
pins, and CCW and FULL are
invoked with a low logic level.
The L6208PD CONTROL pin
deals with decay modes. Decay
modes are defined by the
current paths available within
the H-bridges during MOSFET
switch-off time. A logical low on
the CONTROL pin selects Fast
Decay Mode. All of the H-bridge
MOSFET switches are turned
off for Fast Decay Mode. For
this example, lets work with
MOSFET switches A and D.
Take a look at Figure 3,
which depicts the 1 s dead time
current flow for both Fast and
Slow Decay Modes. During on
time, MOSFET switches A and D
are on providing a current path
through the stepper motor coil.
(It should be pretty obvious that
the direction of the current
flowing through the stepper
motor coil is determined by
which set of opposing MOSFET
switches are energized.) At the
beginning of off time, MOSFET
switches A and D are turned off
and as you can see in Figure 3,
the current flows through the
freewheeling diodes of MOSFET
switch B and MOSFET switch C
by way of the stepper motor coil.
In this scenario, the current
decays very quickly as the volt-
age across the coil is very close to
the power supply voltage level.
When the dead time period
expires, the B MOSFET switch is
turned on and the H-bridge
enters what is called Quasi-
Synchronous Rectification Mode.
Slow Decay Mode is select-
ed by driving the L6208PDs
CONTROL line logically high.
Just as with the beginning of the Fast
Decay Mode cycle, MOSFET switches A
and D are on, allowing current to flow
through the stepper motor coil. At the
beginning of off time, instead of
turning all of the MOSFET switches off,
MOSFET switch A is turned on to open
a current path that includes MOSFET
switch A, the stepper motor coil, and
the freewheeling diode of MOSFET
SERVO 02.2008 51
R
1
1
0
0
C
2
.
2
2
u
F
R
7
0
.
2
0
+
C
3
1
0
0
u
F
R
6
0
.
2
0
B
R
E
S
E
T
C
4
5
6
0
0
p
F
C
1
.
0
1
u
F
R
2
1
0
0
K
R
3
1
0
K
C
5
.
0
1
u
F
D
1
B
A
T
4
6
S
W
S
I
G
N
A
L

G
R
O
U
N
D
C
8
.
0
0
1
u
F
H
A
L
F
/
F
U
L
L
C
O
N
T
R
O
L
C
7
.
0
0
1
u
F
R
5
3
9
K
R
4
3
9
K
+
5
V
D
C
P
O
W
E
R

G
R
O
U
N
D
*
E
N
C
9
.
2
2
u
F
C
W
/
C
C
W
C
L
O
C
K
M
B
I
P
O
L
A
R

S
T
E
P
P
E
R

M
O
T
O
R
A
U
1
L
6
2
0
8
P
D
1
36
18
19
2
7
12
1
1
1
5
2
6
1
3 9
2
4
30
5
25
33
4
2
8 8
3
2
1
0
7
2
2
2
9
GND
GND
GND
GND
H
A
L
F
/
F
U
L
L
SENSEA
C
W
/
C
C
W
O
U
T
1
A
V
R
E
F
B
R
C
A
V
R
E
F
A
R
C
B
VBOOT
O
U
T
2
A
SENSEB
VSB
VSA
C
O
N
T
R
O
L
R
E
S
E
T
O
U
T
2
B
C
L
O
C
K
VCP
O
U
T
1
B
E
N
V
s
SCHEMATIC 1
Eady.qxd 1/10/2008 6:24 PM Page 51
52 SERVO 02.2008
switch C. The decay is slower here as
the voltage across the stepper motor
coil is very low. When dead time
expires in Slow Decay Mode, the C
MOSFET switch is turned on and the
H-bridge enters Synchronous Mode.
I was reluctant to include the
L6208PD EN (Chip Enable) pin in with
the direct logic interfaces we just
talked about. When a logical low is
applied to the EN pin, all of the
L6208PDs power MOSFETs are turned
off. The EN pin is also connected to the
drain of the overcurrent and thermal
protection MOSFET.
In an overcurrent situation, the EN
pin is pulled low by the MOSFET it is
attached to. The forced low on the EN
pin results in the power MOSFETs in both
H-bridges being turned off. So, we have
to be careful how we drive the EN pin.
We can drive EN with an open
collector or push-pull scheme. Since the
PIC I/O ports are predominately push-
pull and not open collector, we will
drive the L6208PDs EN pin in a push-
pull manner as shown in the schematic.
VS
A
and VS
B
are both connected
to the bulk supply voltage V
S
. V
S
must
lie between 8.0V and 52.0V DC. Note
that in Figure 3 the upper MOSFET
switches are N-channel devices. Thus,
the gate drive voltage of the upper
MOSFET switches needs to be greater
than the power supply voltage. The
MOSFET gate drive voltage is
generated with the help of a built-in
charge pump. The L6208PD contains
an internal oscillator whose output is
directly tied to the L6208PDs VCP pin.
Typically, the VCP pin switches
between ground and 10V at a rate of
600 kHz. When the oscillator is at
ground potential, capacitor C1 uses V
S
and the path through diode D1-B to
charge itself. D1-B is reverse-biased
when the charge pump clock
transitions to 10V and the charge that
C1 has accumulated is transferred to
polyester capacitor C9 by way of D1-A.
After a few cycles of this charging
process, the voltage at the VBOOT pin
rises to V
S
+ 10V minus the combined
voltage drops of D1-A and D1-B. The
resultant voltage on the VBOOT pin is
used to drive the H-bridges high-side
MOSFET gates. If youre wondering
where the L6208PD logic power pin is,
there isnt one. The L6208PD has its
own internal 5V and 10V voltage
regulators. All of the L6208PD voltage
rails are derived from V
S
.
The H-bridges contained within the
L6208PD each have their very own PWM
(pulse width modulation) current con-
troller. The sense resistors (SENSEA/R6
and SENSEB/R7) are connected to one
pin of the H-bridge sense comparators.
The other sense comparator inputs are
connected to the VREFA and VREFB pins.
Since the PWM current controller circuit-
ry is identical for both A and B H-bridges,
Ill speak to the PWM current controller
operation in terms of the A circuitry.
The A H-bridge current is sensed by
the A sense comparator, which is using
one of its pins to monitor the voltage
drop across the SENSEA sense resistor,
which is resting between the lower
MOSFET switches and ground. As you
would imagine, as the motor current
increases, the voltage drop across the
sense resistor attached to the SENSEA
pin increases proportionally. The A H-
bridge is turned off when the voltage
across the sense resistor rises above the
voltage that is being applied to the
VREFA pin. The A H-bridge is turned off
by the output of an off-time monostable,
which is triggered by the A sense
comparator when the A sense voltage
exceeds the VREFA voltage. Thus, the
voltage at the VREFA pin and the value of
the sense resistor combine to determine
the maximum current that the A H-bridge
can deliver to the stepper motor coil.
The time that the A H-bridge is
off is determined by the components
hanging off of the L6208PDs RCA pin.
During the off time, the motor current
will recirculate in either Fast or Slow
Decay Mode, depending on which decay
mode has been selected at that time.
The off-time monostable will time out
and turn the A H-bridge back on and the
cycle repeats itself as long as the stepper
motor is attempting to turn its rotor. The
time that the A H-bridge remains in the
off condition is dependent upon the off-
time monostable off time plus the dead
time that occurs between switching the
power MOSFETs on and off.
We are dealing with inductors
(stepper motor coils) here and as you
know, inductors like to throw current
around like a dolphin pitches fish. So,
to keep high reverse currents from
generating sense voltages that may
retrigger the off-time monostable
when the H-bridge is switched back on,
the L6208PD employs the services of a
1 s blanking-time monostable that
prevents the sense comparator from
STMicroelectronics
www.stmicro.com
L6208PD
Microchip
www.microchip.com
PIC18F2620
Sources
Vs
A
FAST DECAY MODE
A B C D = OFF
B D D
C
B
Vs
A
SLOW DECAY MODE
A = ON
B C D = OFF
C
FIGURE 3. In both cases here, MOSFETs A and D were turned on to allow current to
flow through the stepper motor coil during on time.
Eady.qxd 1/10/2008 11:44 AM Page 52
prematurely retriggering the off-time
monostable. The off-time monostable
retrigger range lies between 6.6 s
and 6.0 ms.
That pretty much covers what we
have to contribute to the L6208PD
design as far as a componentry goes.
However, we still have to consider the
microcontroller interface, as well as the
microcontroller firmware build.
So, Whats Next, Fred?
Ive done some reconnoitering.
From what I can make out, we are
going to need a couple of PWM signals
to drive the VREF pins, an ISR (Interrupt
Service Routine) to feed the CLOCK
pin, and enough I/O to drive the
L6208PDs direct interface logic inputs.
If you take another look at the
L6208PD schematic, youll see that I
have placed a voltage divider in the
guise of a potentiometer on the VREFA
and VREFB input pins. That allows us to
use the native full and half step modes
and adjust the stepper motor coil
current by simply twisting on the pot.
I have big plans to implement
microstepping next month and well
replace the potentiometer with a pair
of lowpass filters, which will convert
the PICs modulated digital PWM
signals into analog signals to drive the
VREFA and VREFB pins in microstep-
ping mode. My gut feeling is that we
can get away with a PIC18F2620
microcontroller as it has a couple of
PWM outputs, a fast internal oscillator,
and ample I/O to service the L6208PD.
Ive also determined that we can put
the whole thing down on an inexpen-
sive, double-sided printed circuit board.
So, Im going to get to work on
getting the prototype printed circuit
board ready and Ill also put some
L6208PD rotor-spinning code together
for next time. Now that we have a basic
understanding of what the L6208PD is
and what it can do, well hit the ground
running next month and put all of this
theory into practice. See you then. SV
Fred Eady can be reached via email
at fred@edtp.com
Contact the Author
The
Power SchmartModule
will power up your
circuits with your choice
of -9, -12, +2.5, +3.3,
+5, +9, and +12 volts.
$15
Very Schmart!
www.schmartboard.com
The Escape
Robot`s built-in
microprocessor
enables it to
'think on its own.
(KSR4) $29.95
20 second voice recorder/playback
module. The electret microphone is on the
board. One button records, the other button is
momentarily pressed to replay the message.
(pre-assembled) (A96010) $6.60
5mm White
water clear
LED 3.5V
10,000 mcd
(AB287)
$0.56
The Velleman Personal Scope
is not a graphical multimeter
but a complete portable
oscilloscope at the size and
cost oI a good multimeter.
(HPS10)
The robot Irog moves Iorward
when it detects sound and
repeats: start (move Iorward) -~
stop -~ leIt turn -~ stop -~ right
turn -~ stop. (KSR2) $19.95
S
o
l
d
e
r
i
n
g

r
e
q
u
i
r
e
d
Thousands
m
ore item
s on
our w
eb site!
$146.
Ask for our FREE 96 page catalog
VISIT OUR ONLINE STORE AT
www.allelectronics.com
WALL TRANSFORMERS, ALARMS,
FUSES, CABLE TIES, RELAYS, OPTO
ELECTRONICS, KNOBS, VIDEO
ACCESSORIES, SIRENS, SOLDER
ACCESSORIES, MOTORS, DIODES,
HEAT SINKS, CAPACITORS, CHOKES,
TOOLS, FASTENERS, TERMINAL
STRIPS, CRIMP CONNECTORS,
L.E.D.S., DISPLAYS, FANS, BREAD-
BOARDS, RESISTORS, SOLAR CELLS,
BUZZERS, BATTERIES, MAGNETS,
CAMERAS, DC-DC CONVERTERS,
HEADPHONES, LAMPS, PANEL
METERS, SWITCHES, SPEAKERS,
PELTIER DEVICES, and much more....
ORDER TOLL FREE
1 - 8 0 0 - 8 2 6 - 5 4 3 2
THOUSANDS OF ELECTRONIC
PA RT S A N D S U P P L I E S
Robotics Showcase Robotics Showcase
SERVO 02.2008 53
Eady.qxd 1/10/2008 11:45 AM Page 53
54 SERVO 02.2008
The Key to Battery Runtime
A look at emerging rapid-test technologies for
deep-cycle, lead-acid batteries
CAPACITY:
T
he secret of battery runtime lies
in the capacity. Capacity defines
the energy a battery can hold.
The definition for capacity is usually
given in ampere-hours (Ah); it specifies
the elapsed time when discharging a
battery at a calibrated current to
the end-of-discharge voltage. Portable
batteries commonly use a one hour
discharge; larger batteries are rated at
either a five or 20 hour discharge period.
Lead-acid batteries come in two
basic architectures: deep cycle and
starter types. The deep cycle battery is
designed for maximum capacity and
high cycle count. This is achieved by
installing thick lead plates. Typical
applications are golf carts, wheelchairs,
people movers, scissor lifts, and
inverter power for RVs. In comparison,
starter batteries are made for
maximum CCA (cold cranking amps).
The battery maker obtains this by
adding extra plates to get a large
surface area for maximum conductivity.
Capacity and deep cycling are less
important for automotive because the
battery is being recharged while driving.
If continuously cycled, the thin lead
plates of the starter battery would wear
down rather quickly. As a rule of thumb,
the heavier the battery, the more lead it
contains and the longer it will last.
What is the
Difference Between
Capacity and CCA?
The characteristics of the lead-acid
battery can best be explained by
making capacity responsible for energy
and CCA for delivery. Capacity and
CCA do not age at the same pace. The
CCA tends to stay high through most
of the batterys life, and then drops
quickly towards the end. This often
leaves us stranded when all of a sud-
den the car wont start in the morning.
In comparison, capacity decreases
gradually. A new battery is designed to
deliver 100% of its rated capacity. As
the battery ages, the capacity steadily
drops and it should be replaced when
the reading falls below 70%. Capacity
measurement is considered a more reli-
able state-of-health indicator than CCA.
Lets look at the aging mechanism
of capacity and CCA. Figure 1 shows
two lead-acid batteries: one with high
capacity and one that has aged. The
build-up of so-called rock content as
part of aging robs the battery of usable
energy although it may still provide
good cranking power. Figure 2 illustrates
a battery with high and low CCA by sim-
ulating free-flowing and restricted taps.
The third criterion of battery
runtime is state-of-charge (SoC). The
battery capacity is always measured on
a fully charged battery and the most
simplistic method of estimating SoC is
reading the open terminal voltage
(OTV). This approach is accurate if the
battery has rested for at least four hours
after charge or after applying a load.
The rather long rest period is the
required recovery time to pacify a
battery when disturbed. You should
also be aware that different plate
compositions alter the OTV reading.
Calcium raises the voltage by 5-8%,
affecting SoC estimation. Calcium is
an additive that helps in making the
battery maintenance-free.
Battery Rapid-Test
Methods
Battery capacity is commonly
measured by applying a full discharge.
While this method provides accurate
readings, it is cumbersome, time
consuming, and wears the battery
down unnecessarily. During the last 15
years, several rapid-test methods have
emerged that eliminate the need for
discharge (so the manufacturers claim).
Introduced in 1992, AC conductance
became a popular in testing scheme,
from which CCA is estimated. This non-
invasive method was hailed as a major
breakthrough because the test only
takes a few seconds and the instrument
by Isidor Buchmann
A B
FIGURE 1. Battery capacity illustrated in available liquid. Both batteries are fully charged,
but rock-content limits the capacity on the second battery. A) New battery has high
capacity; B) Aged battery has low capacity.
Buchmann.qxd 1/10/2008 9:46 AM Page 54
stays cool. Unfortunately, AC conduc-
tance is unable to read capacity and is
of limited use for deep cycle batteries.
During the last five years, critical
progress has been made towards
capacity estimations. Cadex has
developed a battery rapid-tester based
on multi-model electrochemical imped-
ance spectroscopy (Spectro). The
Spectro CA-12 injects 24 frequencies
ranging from 20-2,000 Hz. The signals
are regulated at 10 mV to stay within
the thermal battery voltage of lead-
acid. The 24 slices from the frequency
excitations are compared and the
minute nuances analyzed. The instru-
ment completes 40 million transactions
during the short 15-second test.
Electrochemical impedance spec-
troscopy (EIS) is not new. Equipment
using this technology has been in use
for decades. A full-fledged EIS requires
dedicated instruments and a computer
to analyze the data. The set-up is expen-
sive, requires trained staff for analysis,
and is so large that the machinery is
moved on wheels. Furthermore, long
calculation times make the system
unsuitable for commercial use. The
Spectro CA-12 has solved these
problems by using powerful digital
signal processors, but the heart of the
engine lies in the patented algorithm.
What are Typical
Battery Problems?
Lets look at the most common
battery problems and evaluate how
modern battery rapid-testers can
detect these deficiencies. One can
immediately see the benefit of
knowing the capacity.
Low Charge. A low charge reduces
the drive power and the battery
appears weak. Checking a low-charge
battery with a discharge unit will show
low capacity. Rapid-testers such as the
Spectro CA-12 are able to measure the
capacity with a SoC as low as 40%. If
lower, the instrument will prompt the
user to charge and retest (see Figure 4).
Low Capacity. A low capacity battery
will likely have good conductivity and
strong torque. The voltage checks out
fine and everything appears normal,
except the short runtime. Knowing the
capacity on an aging deep cycle
battery is very important because its
the best indication when a battery
should be replaced (see Figure 5).
Mismatched Set. Batteries do not age
at an equal pace. Like the links of a chain,
the battery with the lowest capacity
will govern the runtime. Battery testers
reading capacity can identify low per-
formers and allow a timely replacement.
The high performers can be regrouped
for continued service (see Figure 6).
As encouraging as battery rapid
testing may be, the reader needs to be
reminded that rapid-testers (such as
the Spectro CA-12) are not universal
instruments capable of measuring the
capacity of any battery that will come
along; they need a battery-specific
matrix as a reference. On purchase of
such a unit, the instrument includes
one or several matrices that are
automatically matched with the select-
ed battery. Cadex is in the process of
expanding the matrix library to eventu-
ally include all major battery types.
In time, measuring battery perform-
ance through non-invasive means will
become the acceptable standard, making
discharge methods redundant. Typical
applications are: checking batteries to
reduce false warranty returns; preventing
SERVO 02.2008 55
A B
FIGURE 4
FIGURE 5
FIGURE 2. CCA is shown with free-flowing and restricted taps. CCA is responsible for
starting an engine. Cranking requires 200A; a golf cart typically draws 56A. A) Battery
with high CCA; B) Battery with low CCA.
FIGURE 3. Spectro CA-12. This compact
battery rapid tester measures capacity,
conductance, and state-of-charge in 15
seconds.
unexpected downtime by assessing bat-
tery state-of-health before a breakdown
occurs; and improving the reliability of
battery operated rental equipment.
Designers of battery rapid-test
methods tend to be overly optimistic and
create targets that may not be achiev-
able outside the laboratory. However,
multi-model electrochemical impedance
spectroscopy represents a great leap
forward and opens the door to an
entirely new way of battery testing. SV
FIGURE 6
Isidor Buchmann, founder and CEO of
Cadex Electronics, Inc., has studied the behavior
of rechargeable batteries in practical, everyday
applications for two decades. As an award-
winning author of many articles and books on
the subject, Mr. Buchmann has delivered battery-
related technical papers around the world.
About the Author
Electronic Parts & Supplies
Since 1967
www.c-stamp.com
L
O
W

C
O
S
T
!
CONTROLLERS & ROBOT KITS SENSORS DISPLAYS
Free
Expert
Technical
Support
1 (800) 985-AWIT
E
a
s
y

t
o

C
o
n
n
e
c
t
L
O
W

C
O
S
T
!
LOW COST
Professional SW Dev. Tools
Study online, at your own pace and earn your certificate as a...
www.online-tech-training.com/nv
Programmable Logic Controllers
Technician
For the finest in robots, parts, and
services, go to www.servomagazine.com
and click on Robo-Links.
56 SERVO 02.2008
T
hese new electronics would be
something that could plug in so
that no alterations would be done to
the original shell. After all, I would like
to keep him intact since I hope to find
the original brains and properly restore
him someday! As this project progress-
es, it will be documented and present
some cool ideas that can be used on
your own robots. This first article will
cover a modification I made to the
Handy Board to make it safer to use on
large robots and how I enabled it to
run the larger motors on the BOB base.
There are a few other modifications
that are going to be done to the Handy
Board controller but those deal with
power and expansion and will be
addressed later.
For the moment, the new brain for
BOB is going to be an MIT Handy
Board with the optional expansion
board. It is a 68HC11-based controller
designed for experimental mobile
robotics work. Although there are
more powerful boards on the market,
it is still a great little controller for many
projects. The Interactive-C environment
already provided a lot of the required
functionality and it is easy to add on as
needed. (There is already a library of
extra add-on code and examples on
the Handy Board users group.)
Since the Handy Board is normally
used on smaller robots, something
would need to be done to drive BOBs
larger motors. A couple of issues
needed to be addressed to make this
happen. One is the way it handles the
initialization of the motor logic. The
other is the limited ability to drive large
motors. Many of the ideas to get
around these issues will apply to other
controllers besides the Handy Board.
The Issues
The firmware for the Handy Board
and the users program is stored in a
single 32K static RAM chip at U2. It will
only keep its contents as long as there
is power. When the Handy Board is
switched off, this static RAM chip is still
powered by the battery to preserve its
contents. Even though the SRAM chip
I enjoy restoring early personal robots to their
original working condition. Most of my time is spent
working on the HERO robots, but I always have
other projects going on. One robot in particular an
early Androbot BOB prototype was missing all of
his brains and had been sitting for years as an empty
shell. That was just not acceptable! Since the original
electronics would probably never be found, I decided that
I should at least use some of the extra robot parts from past
projects to get BOB rolling again.
Heres a photo of poor old BOB
in the state I found him in.
by Robert Doerr
Reviving an
Androbot BOB
SERVO 02.2008 57
Doerr2.qxd 1/10/2008 12:40 PM Page 57
58 SERVO 02.2008
draws very little energy, it will
eventually drain the battery. When the
power goes away, the contents of
RAM go away too. Bye, bye, firmware!
A way around this particular issue is
to replace the static RAM with either
the Dallas DS1230 NVRAM or the
Simtek STK16C88 NVRAM chip. These
non-volatile RAM (NVRAM) chips will
keep their contents even if external
power is removed.
In order to use the NVRAM reliably
on any board, there are important
points to consider. I used a DS1230Y-
100 part which has a threshold for
powering down (and storing in NVRAM)
when the supply voltage gets to about
4.5V. The DS1233-10 (U12) RESET chip
used on the Handy Board will hold the
processor in a RESET state if the
voltage is 4.5V or below. Now, say I
had used a DS1230AB-100 part that
has a threshold for storing its contents
at 4.75V. See the potential issue here?
The original DS1233-10 will let
the CPU start as soon as the power
reaches 4.5V but the
NVRAM may not
work yet since it
doesnt become
active until the supply
voltage reaches a
threshold it is
comfortable with.
Switching the RESET
chip to a DS1233-5 which keeps the
RESET until 4.75V would then allow
it to work. Even when the voltage
selected matches, care must be taken
when looking at the timing of these
devices. Some NVRAM chips can still
be disabled for up to an additional 125
ms or more after the supply voltage
reaches a safe level.
If the RESET chip and memory
reference the same threshold, then it
still may not work if the RESET chip
allows the CPU to start accessing the
NVRAM before its ready. To play it
safe, I used the DS1233-5 with the
DS1230Y NVRAM so that I am sure the
NVRAM is ready to go when needed.
There is an application note (AN202)
on the MAXIM website about this
particular issue. It could affect just
about any microcontroller using a
RESET chip and NVRAM together.
This is an important point since
Ive seen quite a few places selling the
DS1233 RESET chip but many dont
mention which version they supply.
Since this may be an issue, make sure
you ask to confirm what flavor of the
DS1233 youre going to get. The origi-
nal RESET chip works fine as-is when
using SRAM on the Handy Board.
When you start altering any board
like this, youll need to look at it
carefully to determine if any other
components should be changed, as
well. The interaction of all the
components you use in a system
should be evaluated to make sure they
all play nice together!
With the NVRAM in place, there is
no longer a need for the secondary 5V
regulator at U14. This can be removed
and a jumper placed from pin 1 of U13
to pin 1 at U14 where the small voltage
regulator was removed. This will
prolong battery life since the memory
power is off along with the rest of
the board. It will no longer drain the
battery when switched off.
There is, however, a bigger issue
with the Handy Board where the
outputs for the motor drivers can come
up in an unknown state when the
firmware is lost. This isnt a huge
problem for small robots where the
Handy Board is normally found. When
the board is powered up, the 74HC374
latch at U8 may not be cleared until the
firmware resets it. This latch controls
the two dual H-bridge drivers that
control up to four motors. If the
firmware is loaded, this isnt a big deal
since it will reset this latch right away.
On the other hand, if the Handy Board
firmware (stored in the battery
backed 32K static memory) isnt
loaded, the latch may not be reset to a
known state.
As a result, your robot could start
rolling away, or worse, try and run you
down when power is applied. Ive
witnessed this behavior firsthand.
Luckily, there is a very simple
modification which will address this
problem and once done, your robot
shouldnt even budge unless you pro-
gram it to do so. It only took swapping
out a single IC chip, cutting one trace,
and adding a jumper so that the
outputs would be cleared whenever
the processor was RESET.
An Easy Modification
NOTE: Static electricity can
damage the CMOS components on the
Handy Board so handle them carefully!
To fix the issue, first replace the
74HC374 in U8 with a 74HC273. On
the back of the Handy Board, the
ground trace going to pin 1 on U8
should be cut. This trace always pulls
the /OE of the 74HC374 low which
forces the outputs of the 74HC374 to
be enabled. The outputs of the
74HC273 are always enabled and this
pin is defined as a reset to clear the
outputs and put them in a known
state. It is an active low just like the
normal RESET logic on the Handy
REVIVING AN ANDROBOT BOB
Modification to the
Handy Board to use
the 74HC273 chip.
Stacked SN754410 chips with heatsink.
Doerr2.qxd 1/10/2008 12:42 PM Page 58
Board. (Dont you just love it when
things fall into place!)
A small jumper is then soldered on
the back of the board from pin 1 of U8
to the output of the RESET chip. This
signal is available at the via just to the
left of R2 (looking from the back). This
will let the built-in RESET logic clear the
contents of the 74HC273 latch at U8
on power-up so all motors will be OFF.
Since I made this modification on the
Handy Board, I have never seen the
motors start by themselves. Ive done
this simple procedure on several Handy
Boards and it appears to have resolved
the issue.
Making it Better:
Plug-in Upgrades
There are a lot of factors to
consider when sizing an H-bridge to a
particular motor. This topic alone could
warrant several articles and there are
already numerous references on the
web about this. Let me just say that it
is usually best to err on the side of
caution and select something a good
margin higher than what you expect
your motor/drive system will call for.
You may have sized things perfect
when there isnt a lot of load on your
drive system, but you might see a
catastrophic failure (complete with
magic smoke) if a motor stalls or your
robot gets stuck. You can certainly
expect bad things to occur if you are
going to exceed either the current
capacity or working voltage of the
original circuit. Plan for the worst case
and you shouldnt get many surprises.
Right from the start, I knew the
built-in motor controller chips (L293D
or SN754410) could not handle the
current required to drive the two main
drive motors in the Androbot base. I
wanted to use the built-in Handy Board
code to drive the base since the
commands were already present within
Interactive-C. To keep it clean, I wanted
something I could just plug in and use.
This called for the design of a plug-in
upgrade to replace the H-bridge chip
on the Handy Board. The results should
also work with other controllers
such as the Handy Cricket, SUMO11, or
devices that use the L293D or
SN754410 H-bridge driver chips.
To get a little more capacity for
any boards that use the L293D
H-bridge chip, simply swap the L293D
with a SN754410 chip. This will
increase current capacity from 600 mA
to 1A (2A peak). You cant get any
easier than just replacing a chip that
works if you are just starting to exceed
the ratings of the L293D.
This wasnt an option for my
project since the large main drive
motors in the Androbot base would
easily fry these driver chips. I needed a
more powerful option.
NOTE: Many Handy Boards now
ship with the SN754410 chips since
they have a higher rating and are more
robust than the L293D chips.
The next step is also an easy one.
You can piggyback two SN754410 to
increase current capacity to nearly 2A
(4A peak). It is recommended to use a
DIP heatsink on the top chip with a bit
of heatsink grease on both chips. This
will allow the heatsink to cool the
bottom chip, as well at the top one.
The leads on the two chips are then
soldered together so that the pair can
be installed in place of the original
L293D or SN754410 chip.
This, unfortunately, was still not
nearly enough for the main drive
motors I wanted to use in the
Androbot base. They may make the
difference in your own project, though.
When going this route, check to make
sure there is physical clearance for
the extra height of the second chip
and heatsink.
Dedicated H-bridge Board
To take it to the next level, I made
a small board that would act like a
plug-in L293D chip. It connects with a
short DIP ribbon cable to drive a larger
H-bridge. The first iteration used a
L298N to increase the total capacity to
Custom L298 H-bridge board.
L298 H-bridge board plugged into the
Handy Board (very clean look).
Dual LMD18200 H-bridge board.
Custom H-bridge board for use
with LEAF robots.
SERVO 02.2008 59
Doerr2.qxd 1/10/2008 12:43 PM Page 59
60 SERVO 02.2008
4A up to 46V. Since these chips have
similar features, this made sense to
start with. The physical layout of the
components is similar to one made
several years ago that is posted on the
DPRG website. However, I wanted to
make one specifically for use with the
Handy Board that would just plug in.
The circuit closely follows the
schematics set forth in the L298N
datasheet with the exception of the
connections to the host. These
connections are mapped out to a
16-pin DIP to match the smaller DIP
H-bridge chip. It just took mapping out
the appropriate pins for the L298N to
match up with those of the original
L293D. Since the L298N lacks the
internal protection diodes present
within the L293D, those had to be
added externally on the board.
Fast rectifier diodes were selected
for better protection. The L298N is
mounted securely to a heatsink to aid
in cooling and also protect it physically.
This board has large connectors for a
separate power connection
and for both motors. These
power connections have
.156 spacing to accept a
two-pin header or two-pin
screw terminal. Its nice to
have the choice for
installing either style
connector. There are also
pads for the current sense
signal so that can be used later.
An extra set of pads were added
to feed the motor power back through
the 16-pin DIP socket, if desired. The
pads are a bit unique as either surface-
mount or older axial devices can be
used. I just call them unipads since they
support either style package. Most
applications will just leave these open.
The result was a slick little H-bridge
driver board that can just plug in place
of a L293D chip. Since the DIP
connector on the ribbon cable is about
the height of the IC, it can easily be
used on the Handy Board with an
expansion board installed. It can work
in places where a piggybacked set of
SN754410 chips will not fit physically.
At first I thought this might work
for the main drive motors on the BOB
base, but during some aggressive test-
ing at 24V with both motors running it
just gave up. At least I didnt set off the
fire alarm! I had underestimated the
load on the motors a bit. This new
driver board was easily large enough to
move the head motor (once
the chip was replaced, of
course), but something a bit
beefier was needed for the
main drive motors.
Another Step Up
Since I really liked the
idea of a separate H-bridge
board that would plug in
place of the L293D, I made
another design. This time, I used a pair
of LMD18200T chips to increase
capacity to 2x3A (up to 55V) with the
same 16-pin DIP cable as a plug-in
replacement module. The LMD18200T
is a popular H-bridge chip and is easy to
work with. This required mapping out
the L293D to a pair of LMD18200T
chips as was previously done for the
L298N board. Since the LMD18200T
already contains internal protection
diodes, it simplified the board layout.
This is the most robust of the driver
combinations presented here. It will
provide up to 3A (6A peak) per motor
up to 55V.
This board has heavy duty
heatsinks which improve the reliability
of the LMD18200s, as well as provide
secure mounting for them. These
H-bridge chips also have a few special
features such as braking and current
sense. A SIP header was installed on
the board to enable use of these extra
signals. This combination works well
for driving the BOB base at 24V. No
smoke! So far, this driver board has
performed fantastic and Ive had BOB
moving all over the place.
One other very handy feature on
this particular board is that if either
motor happens to be running in the
wrong direction, all you have to do is
move a jumper (J1 and J2) to swap
directions. No re-wiring of your robot is
required. This feature works well with
the Handy Board and should also work
on other controllers that use the L293D
in a similar fashion. It all depends on
what signals they supply to each one of
the L293D half H-bridges. Normally, the
signals to each half H-bridge in the
L293D are inverted and that is what
the jumper leverages. If the motor goes
the wrong way, moving the jumper just
picks up the inverted signal.
This last H-bridge board worked
out so well that a slightly different
variant was created with both .100
SIP headers and four-pin modular ends
for the inputs. It is a plug-and-play ver-
sion that will work with LEAF-based
robots and plugs directly into the LEAF
controller. Ive started to build a LEAF
base for another robot and am going
to use this H-bridge to drive it. This
version may make it easier for those
Both custom H-bridge
boards in use. The larger
dual controller for the
main drive and the smaller
one for the head.
Dual LMD18200 board in
place to drive BOBs motors.
Doerr2.qxd 1/10/2008 12:44 PM Page 60
who also intend to build a LEAF robot.
For those that would like to build
one of these enhanced LMD18200
plug-in H-bridge boards, blank PCBs
(and possibly kits) will be available from
the author.
Conclusion
A couple of simple modifications
can really improve an existing robot
controller like the Handy Board and
make it more suitable for use in larger
robots. Fixing the issue with the
controller to ensure the motors are off after a reset was
critical so the robot doesnt run you over. The new H-bridge
boards presented here can easily upgrade controllers or
devices using the L293D/SN754410 chips. This will allow
them to drive larger motors without any other alterations to
the software or hardware. They just plug in. You cant get
any easier than that!
Now that the Handy Board can move BOB around, the
next topics are: work on the power distribution/charging
system; use some DIP relays to multiplex a Polaroid sonar
module so that it can drive five transducers; add a voice using
the SpeakJet chip; add a co-processor to control the head
and lights; get the encoders going again for the drive system;
and connect a few more sensors so that BOB can sense and
interact more with the world around him. Many of these
have already been finished and will appear in future articles.
I guess, until the original innards turn up, hell just be
nicknamed Handy BOB ...
Keep those old robots alive! SV
Robert Doerr can be reached via email at rdoerr@bizserve.com.
RobotWorkshop Creator of the
plug-in H-bridge boards for the L293D
www.robotworkshop.com
Main Handy Board site (also Handy Cricket links)
www.handyboard.com
SUMO11 Board (Similar to Handy Board and uses
same H-bridge chip)
www.1sorc.com/
Main website for the LEAF project.
www.leafproject.org
Application Note on using L293 and L298 devices
www.st.com/stonline/books/pdf/docs/1681.pdf
Datasheet for L293D
Datasheet for SN75441
http://focus.ti.com/docs/prod/folders/print/sn754410.html
Datasheet for L298N
Datasheet for the LMD18200
www.national.com/ds/LM/LMD18200.pdf
Maxim IC Application Note AN202
http://pdfserv.maxim-ic.com/en/an/app202.pdf
DPRG L298 H-bridge project similar to the L298 board
referenced here
www.dprg.org/projects/1998-04a/
Web References
L293D - 600 mA (1.2A peak) up to 36V - DIP 16
Quad Half H-bridge
SN754410 - 1A (2A peak) up to 36V - DIP 16
Quad Half H-bridge
L298N - 4A (6A peak) up to 46V - Multiwatt 15
Dual Full H-bridge
LMD18200 - 3A (6A peak) up to 55V - Multiwatt 11
Dual Full H-bridge
Listing of common H-bridge driver chips and their ratings
SERVO 02.2008 61
Doerr2.qxd 1/10/2008 12:45 PM Page 61
62 SERVO 02.2008
T
heres an old saying maybe its a
new saying that goes Everything
old is new again. What we used to
throw out in the trash are now
collectors items people buy and sell on
eBay for top dollar. Popular fashions
that peaked then ebbed in decades
long past and were considered pass,
are now all the rage. Again.
Recycling is something we humans
do by nature. Old things get repur-
posed inside new ideas. In the case of
mechanical constructions such as
robots, recycling is often cheaper and
usually faster. If you already have it on
hand, it doesnt cost you any extra, and
its available for your immediate use.
Recycled robots is itself a recycled
idea. Theres nothing new in the
concept of constructing robots from
used, surplus pieces. What is new are
the vast resources now available to find
those pieces. Aside from the usual
haunts such as local thrift stores
and the garbage bins behind the
neighborhood light industrial parks,
we now have the Internet and its
searching features to locate the best
junk for building recycled bots.
In this months Robotics
Resources, well cover a number of
these used parts hot spots, and provide
useful guidance on how to best fill your
robotics parts bin.
Parts and More Parts
The Internet is a vast mail order
catalog, but mail order isnt the only
(or necessarily the best) way to find
robot bits and pieces. When you buy
local, not only do you get to see
and touch the item(s) youre interested
in, you get to have it right then
and there. And, of course, you save on
shipping costs.
Make your first destination the
nearby thrift stores. Most areas have at
least one Goodwill, Salvation Army,
or similar thrift store that accepts
household discards and resells them at
a fraction of their original price. Look
for: VCRs, record players, CD and DVD
players, fax machines, battery-operated
tools (drills, socket wrenches, etc.), and
most anything else powered by small
DC motors. In addition to one or
more motors, inside youll often find
wiring, gears, pulleys, belts, and
other mechanical parts useful in robot
building.
You dont need to worry if what
youre buying still works; in fact, you
often get the best prices for the old
broken junk. Over the years, Ive
bought dozens of VCRs great
sources for small motors and other
precision parts for a couple of bucks
each. The older the VCR the better,
because it will have more and better
parts to pull out.
Other prime local sources include
garage sales, appliance repair shops
(ask to go through their discard bin),
and dumpsters behind electronic and
mechanical manufacturing businesses.
For the latter, its generally not legal to
raid a garbage bin unless its on public
property, and most of these are on
private property. So get permission
from the owner first. Be careful of
broken glass and other sharp objects in
the trash. Wear heavy gloves or youll
risk getting a serious cut.
Once youve exhausted the
neighborhood venues, you can turn to
mail order and the Internet. See the
Sources section that will follow for a
selected list of online retailers that
specialize in used and surplus mechan-
ical parts. But by no means consider
this list as complete. These initial
sources will help give you a general
idea of whats out there and what
might be useful for the robots you
build. Note what things are called and
the general asking prices. Then broad-
en your search using Internet engines
such as Google, Yahoo, or MSN.
Remember that there isnt an
endless supply of used and surplus
merchandise. Quite often, the seller
has only hundreds or perhaps just
dozens of a particular item. The odds
of you looking and finding just the
right part when the seller has them for
sale are not high, so you need to plan
ahead and purchase items for future
projects. The danger in this approach is
getting too excited about all the good-
ies out there, and spending money on
things youll never use. You should limit
buying for the parts bin as opposed
to making purchases for your current
project to only low-cost items.
One approach to spreading out
the cost of filling a parts bin is to get
involved with friends. Make group
purchases for a centralized parts bin
that you all have access to. Keep track
of the cost of each item, including a
pro-rated amount for shipping. As
items are taken out by the members of
The Recycled Robot
Tune in each month for a heads-up on
where to get all of your robotics
resources for the best prices!
RoboResources.qxd 1/10/2008 9:57 AM Page 62
your group, have them repay the
purchase price. The money is put back
into an account, and is used for the
next mail order purchase.
Even with careful planning, youll
find your parts bin will eventually
overflow with bits of junk you end up
not needing. You can resell these on
eBay, horse trade them with fellow
robot builders, drop them off at the
recyclers, or give them away. Nothing
should ever be thrown into the trash,
especially if it contains recyclable
content, such as copper wire.
Designing Around
Found Parts
The challenge of using surplus
parts is that you dont always find
exactly what you need. To effectively
build robots out of recycled compo-
nents, you need to be ready to adapt
to the parts on hand.
Case in point: motors from a VCR.
The typical VCR contains more than one
motor, but in most cases, each motor is
different. And these motors tend not to
have gear boxes attached to them,
which would make them ideal for use as
drive motors on a small robot. This
makes designing and constructing the
typical differentially steered robot (two
motors; a motor and wheel on each
side) a challenge. Unless youre lucky
enough to find two identical used VCRs,
you will only have one of each kind of
motor. Differentially steered bots
require two motors of the same type
and specifications.
Instead, you need to adapt to the
parts by planning a different type of
robot. Instead of differentially steered,
perhaps you could build a car type
robot, with one main drive motor and
a small steering motor. If you cant find
a suitable motor for precise steering,
you can purchase a hobby servo for
$10-$15 (theres nothing keeping you
from mixing old and new the idea is
to simply stretch your robot-building
dollars by using as much recycled
components as you can).
The same goes for power transmis-
sion components: gears, pulleys, and
belts. Youre not likely to happen on
exactly the right size of gear for your
carefully planned robot project.
Instead, look first in your parts bin, and
then design your bot around what you
have. Youll save time and money, and
no one else will know the difference.
If youre familiar with BEAM
robotics (BEAM stands for Biology,
Electronics, Aesthetics, and Mechanics),
youre already familiar with this
approach. Though not a requirement,
BEAM robots are often made from
surplus or found parts. The diversity
of BEAM designs allows for a wide
variety of applications of these
found parts.
For example, many BEAM projects
diverge from the typical two wheeled
design of the typical desktop bot.
Some may hop on one or two legs,
others may roll on an obtuse arrange-
ment of wheels, and yet others may
swim, crawl, or walk. Often, the myriad
of designs comes from the selection of
available parts on hand. By limiting the
design to whatever is in the parts bin,
new and different types of creations
are often possible.
Keeping Track of
What You Have
For most anyone involved in the
robot building craft, in reality the
parts bin doesnt exist. Its more like
parts bins. The more bots you build
and the longer you stay at it, the more
bins youll amass. And with them, the
chances of forgetting what you already
own and unnecessarily buying things
you dont need.
How you organize your parts is up
to you, but Ive found a system of
transparent small plastic boxes to be
among the best solutions. You can
purchases these boxes at dollar and
discount stores. Avoid the larger boxes
where parts are likely to get lost
among all the others. Shoebox size
boxes have worked best for me; try a
few to see if theyre adequate for your
needs. Being transparent, you dont
need to open each one to locate what
you need, but its still a good idea to
write down the general contents on a
label placed on the side of the box.
I have separate boxes for motors,
wheels, gears, and R/C (radio control)
parts. Because motors tend to be large
and heavy, I have several boxes, each
with a different style of motor. Small
motors go into one box, larger or gear
head motors go into another.
Additional boxes are reserved for
miscellaneous small parts, which are
All Electronics is one of the primary sources for new and used robotics components.
SERVO 02.2008 63
64 SERVO 02.2008
contained in reclosable zipper plastic
bags. You can write notes on the
bags, and arrange them in the boxes
as needed.
I once tried a more elaborate
inventory system consisting of an Excel
worksheet listing the item, price, place
purchased, and a description, but I
found the upkeep to be too time-
consuming. You might want to try it
for the larger, more expensive items.
Such a system might also be useful if
you share your parts bins with others,
such as a user group or school.
Recycling Your
Own Robots
Lastly, consider than your complet-
ed robots are themselves sources for
parts for your next project. If youre on
a tight budget, you might not have the
luxury of keeping all your finished
robots on a shelf. Gutting them and
reusing their parts is an efficient if
not sometimes depressing way of
getting more mileage out of your
investment.
Before I tear apart an old bot, I like
to take pictures of it, for no other
reason than as a sentimental keepsake.
If the robot had some unique locomo-
tion system, I might record a few
minutes of it on the camcorder. With
the visual log made, the parts are sum-
marily pulled off and reused, and the
old robot fades into a distant memory.
Should you wish to reuse parts
from one robot to the next, consider
ease of dismantling your previous
creation. Avoid the use of tapes, glues,
and other adhesives; when possible,
secure things using nuts, bolts, and
other fasteners. Options for easy
removal and reuse include tie-wraps
and heavy-duty Velcro.
Sources
A-2-Z Solutions, Inc.
www.a2z-solutions.com
New and surplus electronics.
Mostly computer equipment (PCs,
monitors, scanners, and so forth).
AE Associates, Inc.
www.ae4electronicparts.com
New and used electronics, includ-
ing switches, connectors, electronic
components (resistors, capacitors,
diodes, transistors, etc.), and test
equipment. Searchable database.
All Electronics Corp.
www.allelectronics.com
All Electronics is one of the primary
sources for new and used robotics
components. Prices and selection are
good. Walk-in stores in the Los Angeles,
CA area. Product line includes motors,
switches, discrete components,
semiconductors, LEDs, infrared and CdS
sensors, batteries, LCDs, kits, and much
more. Specification sheets for many
products are available on the website.
Alltronics
www.alltronics.com
New and surplus merchandise.
Among their product line useful in
robotics are DC and stepper motors,
stepper motor controllers, power
MOSFETs, small CCD video cameras,
and tools.
American Science & Surplus
www.sciplus.com
AS&S sells surplus of all types,
including some youd normally find in
an Army/Navy surplus store. But they
also carry motors, gears, batteries,
switches, and some electronics.
APEX Electronics
www.apexelectronic.com
Military and industrial surplus,
with a major emphasis on wire of all
types and sizes. Also motors, switches,
gears, and other components. Huge
selection in store, with online sales for
popular items.
Apex Jr.
www.apexjr.com
Surplus electronics and mechani-
cals. General electronics, transformers,
and movie props.
B.G. Micro
www.bgmicro.com
A haven for the electronics tinker-
er and robotics enthusiast. Much of the
stock is surplus, so it comes and goes,
but while its being offered, it has a
good price attached to it. Large down-
loadable catalog.
BCD Electro, Inc.
www.bcdelectro.com
Surplus electronics: active and
AS&S sells surplus of all types, but they also carry motors, gears,
batteries, switches, and some electronics.
passive electronics, motors, relays,
switches, etc.
BMI Surplus
www.bmius.com
Electronic surplus, much of it
high-end industrial or scientific;
opticals, and laser.
Burden Surplus Sales
www.surpluscenter.com
Industrial and commercial surplus,
with a heavy emphasis on larger
motors, pneumatics, and hydraulics.
If youre building a big robot, come
here!
C & H Sales
www.candhsales.com
C & H is primarily a source for
mechanical surplus, but they also
stock various passive components, as
well as switches, relays, and motors.
Online sales through Web catalog,
printed catalog, and walk-in store in
Pasadena, CA.
EIO.com
www.eio.com
Surplus sales of all major
electronics components, plus links and
resources.
Electro Mavin
www.mavin.com
Electronic components, motors,
batteries, optics, and test equipment.
Electronic Dimensions
www.el-dim.com
Military and industrial surplus,
electronics, radio receivers, transmit-
ters and parts, electron tubes, test
equipment, and ham gear.
Electronic Goldmine
www.goldmine-elec.com
Electronic Goldmine has new and
used electronic components (LEDs,
potentiometers, resistors, heatsinks,
transistors, etc.), robot items, electron-
ic project kits, and more.
Electronic Surplus, Inc.
www.electronicsurplus.com
Wide selection of test equipment
and electronics parts.
Electronics Plus
www.electronics-plus.com
Product lines include batteries and
battery accessories; cable and wire;
capacitors and resistors; chemicals;
circuit boards and accessories; coils,
chokes, inductors, and more.
Electronix Express
www.elexp.com
New and surplus electronics,
including passive components, motors,
B.G. Micro is a haven for the electronics tinkerer and robotics enthusiast.
Electronic Goldmine sells new and used electronic components,
robot items, electronic project kits, and more.
SERVO 02.2008 65
relays, and more.
Excess Solutions
www.excess-solutions.com
Surplus electronics. Local store and
online sales.
Fair Radio Sales
www.fairradio.com
Though specializing in surplus for
ham radio, Fair Radio also offers plenty
of general electronics and test equip-
ment. Online sales with Web catalog.
Gateway Electronics, Inc.
www.gatewayelex.com
Electronic surplus store in St. Louis,
MO, with some online sales.
HSC Electronic Supply
www.halted.com
Online mail order sales, with walk-in
retail stores in California. Halted offers a
mix of computer and electronics surplus.
Marlin P. Jones & Assoc., Inc.
www.mpja.com
Wide assortment of electronic
components at decent prices. Online
sales through Web catalog, but get the
printed catalog for browsing.
MCM Electronics
www.mcmelectronics.com
New and some surplus electronics.
Offers include audio and video, and serv-
ice components (e.g., belts, gears, pinch
rollers that can be used as wheels, etc.).
MECI Mendelson
Electronics Company, Inc.
www.meci.com
Surplus electronics, motors, and
even a special section for combat robot
parts large motors, batteries, that
sort of thing.
Skycraft Parts & Surplus, Inc.
www.skycraftsurplus.com
Surplus mall offering power
supplies, transistors, relays, ICs, wire,
cable, heat shrink, transformers, motors,
fiber optics, test equipment, resistors,
diodes, and more.
Surplus Sales of Nebraska
www.surplussales.com
Surplus electronic parts, including
connectors, passive and active compo-
nents, electronics hardware, relays and
solenoids, chemicals, and antique radio
parts (to build antique robots, of course).
Timeline, Inc.
www.timeline-inc.com
Surplus of all kinds: electronic,
computer peripheral, laser, motors,
LCDs, and more.
Unicorn Electronics
www.unicornelectronics.com
Large selection of electronic com-
ponents, including passives, transistors,
logic ICs, relays, and more. SV
Gordon McComb can be reached
via email at robots@robotoid.com
CONTACT THE AUTHOR
Whats the difference?
Price!
Youre correct, the one on the right costs
less because its from Jameco! When
youre looking to purchase major name
brand passives or semiconductors, be
sure to check out Jameco first. They also
offer money-saving generic equivalents
at about 20% less, so youll save even
more! And if you should find a lower
print-advertised price for a product
Jameco offers, just let them know and
theyll offer that product for 10% less
than the other companys price, right
down to Jamecos cost! Wouldnt it be
worth your time to check out Jameco
today at www.Jameco.com?
Both Major Name Brands and Generic Products for Additional Savings
Over 99% of Catalog Products are in Stock for Immediate Delivery
Only Jameco Guarantees the Lowest Prices
Click or call today for a FREE Jameco Catalog
& start benefiting from Jamecos Awesome Prices!
Visit www.Jameco.com/NVT
Call 1-800-831-4242
66 SERVO 02.2008
// castling bonuses
B8 castleRates[]={-40,-35,-30,0,5};
//center weighting array to make pieces prefer
//the center of the board during the rating routine
B8 center[]={0,0,1,2,3,3,2,1,0,0};
//directions: orthogonal, diagonal, and left/right
from orthogonal for knight moves
B8 directions[]={-1,1,-10,10,-11,-9,11,9,10,-10,1,-
1};
//direction pointers for each piece (only really for
bishop rook and queen
B8 dirFrom[]={0,0,0,4,0,0};
B8 dirTo[]={0,0,0,8,4,8};
//Good moves from the current search are stored in
this array
//so we can recognize them while searching and make
sure they are tested first
by James Isom
A
bi-m
onthly
colum
n for
kids!
LESSONS
FROM THE
LABORATORY
LESSONS
FROM THE
LABORATORY
NXT Packbot:
Part 3
STEP 3:
STEP 6:
Parts:
Parts:
Parts:
STEP 1:
Parts:
Parts:
STEP 4:
STEP 7:
Parts:
Parts:
STEP 2:
STEP 5:
L
ets pick up where we left off in December
and finish up the main chassis of the Packbot.
Rear Chassis Assembly:
The two sides are mirror images of one another
that come together with a middle bracket that
will eventually help organize the wires for your
motors and sensors. Once again, Ill include
instructions for both sides to make it easier.
Right Side:
SERVO 02.2008 67
LessonsFromTheLab.qxd 1/10/2008 3:47 PM Page 67
68 SERVO 02.2008
STEP 9:
STEP 10:
STEP 13:
Parts:
STEP 8:
STEP 11:
STEP 14:
Parts:
Parts:
STEP 12:
Parts:
Parts:
Parts:
Parts:
Parts:
Note: Due to technical issues, all the treads represented in these instruc-
tions are the older RCX style treads. You should use the new style treads
(theyre less toothy on the inside and stretch more) available in the
Education Resource Set available from LEGO Education.
Left Side:
Parts:
Parts:
STEP 17: STEP 16: STEP 15:
Parts:
Parts:
Parts:
Parts:
STEP 18:
STEP 21:
STEP 24:
STEP 20:
STEP 23:
STEP 26:
STEP 19:
STEP 22:
STEP 25:
Parts:
Parts:
Parts:
Detach the outside beams and slide the rear portion
of the front tread over the wheel and reassemble.
Rear Bracket Assembly: Part 1
SERVO 02.2008 69
70 SERVO 02.2008
STEP 28:
STEP 31:
STEP 33:
STEP 36:
Parts:
Parts:
Parts:
Parts:
Parts:
STEP 27:
STEP 30:
STEP 34:
STEP 37:
STEP 29:
STEP 32:
STEP 35:
Parts:
Parts:
Add the other half of
the tread assembly to
complete the back
portion of the robot.
Your Packbot should
now look like this.
Rear Bracket Assembly: Part 2
Rear Assembly:
Putting it all together
STEP 38:
STEP 41:
STEP 44:
STEP 47:
Lets brace the sides to add
a bit more rigidity. Make
two of the following.
Parts:
Parts:
STEP 40:
STEP 43:
STEP 46: STEP 49:
Parts:
Parts:
Parts:
Parts:
STEP 39:
STEP 42:
STEP 45:
STEP 48:
Parts:
Parts:
Parts:
Parts:
SERVO 02.2008 71
Arm Drive Train Assembly:
Arm Assembly: Right
72 SERVO 02.2008
STEP 50:
STEP 53:
STEP 56:
STEP 59:
Parts:
Parts:
Parts:
Parts:
STEP 51:
STEP 54:
STEP 57:
STEP 60:
Parts:
Parts:
Parts:
STEP 52:
STEP 55:
STEP 58:
STEP 61:
Arm Assembly: Left
Parts:
Parts:
Parts:
Parts:
Were almost finished. Next time, we will add
the NXT, wire it up, and get programming. SV
Robot Builders Cookbook
by Owen Bishop
This is a book for
first-time robot
builders, advanced
builders wanting to
know more about
programming robots,
and students in
further and higher
education tackling
microcontroller-
based practical
work. They will all
find this book a unique and exciting source
of projects, ideas, and techniques to be
combined into a wide range of fascinating
robots. $29.95
123 Robotics Experiments
for the Evil Genius
by Myke Predko
If you enjoy tinkering
in your workshop and
have a fascination for
robotics, youll have
hours of fun working
through the 123
experiments found in
this innovative project
book. More than just
an enjoyable way to
spend time, these
exciting experiments also provide a solid
grounding in robotics, electronics, and
programming. Each experiment builds on
the skills acquired in those before it so
you develop a hands-on, nuts-and-bolts
understanding of robotics from the
ground up. $24.95
FIRST Robots: Aim High
by Vince Wilczynski / Stephanie
Slezycki
This book looks at
30 different robot
designs all based on
the same chassis,
and provides in-
depth information
on the inspiration
and the technology
that went into build-
ing each of them.
Each robot is fea-
tured in 6-8 pages providing readers with a
solid understanding of how the robot was
conceived and built. There are sketches,
interim drawings, and process shots for each
robot. $39.95
Mechanisms and Mechanical
Devices Sourcebook
by Neil Sclater / Nicholas Chironis
The fourth edition
of this invention-
inspiring engineering
resource covers the
past, present, and
future of mechanisms
and mechanical
devices. Youll find
drawings and
descriptions of more
than 2,000 compo-
nents that have
proven themselves over time and can
be incorporated into the very latest
mechanical, electromechanical, and mecha-
tronic products and systems. Overviews of
robotics, rapid prototyping, MEMS, and
nanotechnology, along with tutorial chapters
on the basics of mechanisms and motion
control, will bring you up-to-speed quickly
on these cutting-edge topics. $89.95
We accept VISA, MC, AMEX, and DISCOVER
Prices do not include shipping and
may be subject to change.
CNC Robotics
by Geoff Williams
CNC Robotics gives you
step-by-step, illustrated
directions for designing,
constructing, and testing
a fully functional CNC
robot that saves you 80
percent of the price of an
off-the-shelf bot and
that can be customized
to suit your purposes
exactly, because you designed it. Written
by an accomplished workshop bot
designer/builder, this book gives you
everything you need. $34.95
The SERVO Webstore
The Amateur Scientist 3.0
The Complete Collection
by Bright Science, LLC
There are 1,000 projects
on this CD, not to
mention the additional
technical info and
bonus features. It
doesnt matter if youre
a complete novice
looking to do their first
science fair project or a
super tech-head gadget
freak; there are enough
projects on the single CD-ROM to keep you
and 50 of your friends busy for a lifetime!
$26.99
Attention Subscribers ask about your discount on prices marked with an *
N N
E E
W
!
W
!
Forbidden LEGO
by Ulrik Pilegaard / Mike Dooley
Build the Models
Your Parents
Warned You
Against.
Forbidden LEGO
introduces you to
the type of free-
style building that
LEGOs master
builders do for fun
in the back room. Using LEGO bricks in com-
bination with common household materials
(from rubber bands and glue to plastic
spoons and ping-pong balls) along with
some very unorthodox building techniques,
youll learn to create working models that
LEGO would never endorse. $24.95
SERVO 02.2008 73
WebstoreFeb08.qxd 1/10/2008 12:58 PM Page 73
Get your very own limited-edition SERVO Magazine T-shirt. Shirts come
in sizes S, M, L, and are available in either black or white.
All shirts are 100% preshrunk cotton. Please check availability on our website.
SERVO
Magazine
T-Shirts
For men and women
B
lo
B
low
O
u
t
w
O
u
t
S
p
e
c
ia
l
S
p
e
c
ia
l
$
9
.9
5
!
$
9
.9
5
!
From HomoSapien to RoboSapien Before R2D2 there was R1D1
Take This Stuff and Hack It!
by Dave Prochnow
Transform common household items into
really cool stuff. You
don't need to be an
electronics genius to get
started turning everyday
items into high-performing
wonders. With how-to
guru Dave Prochnow's
step-by-step directions
and fully illustrated plans,
even beginners can hack their way to a high-
tech home, cooler toys, and less yard work.
Certain to fire your imagination and start you
plotting new, original, and even more creative
wonders you can make from ordinary house-
hold items, Take This Stuff and Hack It! is the
perfect gift for your inner inventor. $27.95
Building Robots with LEGO
Mindstorms NXT
by Mario Ferrari, Guilio Ferrari
The Ultimate Tool for
MINDSTORMS
Maniacs, the new
MINDSTORMS kit has
been updated to
include a
programming brick,
USB cable, RJ11-like
cables, motors, and
sensors. This book
updates the robotics information to be
compatible with the new set and to show
how sound, sight, touch, and distance issues
are now dealt with. $39.95
NEED MORE
INFO?
Go where
SERVE goes!
The online stor The online store @ e @
www www.ser .ser v vomagazine omagazine.com .com
Robot Builders Bonanza
Third Edition
by Gordon McComb / Myke Predko
Everybodys favorite
amateur robotics book
is bolder and better
than ever and now
features the fields
grand master Myke
Predko as the new
author! Author duo
McComb and Predko
bring their expertise
to this fully-illustrated robotics bible to
enhance the already incomparable content
on how to build and have a universe of
fun with robots. $27.95
SERVO Magazine
Bundles
Published by T & L Publications, Inc.
To order call 1-800-783-4624
Now you can get one years worth of all
your favorite articles from SERVO Magazine
in a convenient bundle of print copies.
Available for years 04, 05, 06, and 07.
$57
per bundle
S
a
v
e
$
1
0
o
ff th
e
n
o
rm
a
l
p
ric
e
!!
Robot Builders Sourcebook
by Gordon McComb
Fascinated by the
world of robotics but
dont know how to
tap into the incredible
amount of informa-
tion available on the
subject? Clueless as
to locating specific
information on robot-
ics? Want the names,
addresses, phone
numbers, and websites of companies that
can supply the exact part, plan, kit, building
material, programming language, operating
system, computer system, or publication
youve been searching for? Turn to Robot
Builders Sourcebook a unique clearing-
house of information that will open 2,500+
new doors and spark almost as many new
ideas. $24.95
74 SERVO 02.2008
BACK ROOM SPECIALS
Concise Encyclopedia
of Robotics
by Stan Gibilisco
Written by anaward-win-
ning electronics author,
the Concise Encyclopedia
of Robotics delivers 400
up-to-date, easy-to-read
definitions that make even
complex concepts
understandable. Over
150 illustrations make the
information accessible at a
glance and extensive cross-referencing
and a comprehensive bibliography facilitate
further research.
$19.95 Sale Price $9.95 Only 2 Left
Pub date: November 12, 2002
SERVO'S CUSTOM
EMBROIDERED HAT
This is one good looking hat, with its stone
washed blue cap and bright yellow/gold
stitching. There's no better way to show
everyone that you're well endowed with a
bigger - than - average brain then by proudly
displaying your affiliation with SERVO
Magazine. $ 14.95*
Hydrogen Powered Racer and Hydrogen Station
Recently named as one of the Best Inventions of 2006 by Time Magazine, the H-racer is now the
best selling fuel cell product in the world. The H-racer is a micro-version of what engineers and
scientists have been dreaming about for real cars: combining hydrogen with oxygen to generate
a DC current to power an electric motor. Unlike a gas-powered car engine, the only byproducts
of this electrochemical process are electricity, heat, and pure water. With the H-racer, you can
witness the power of new energy technology in the palm of your hand. Horizon has created
this unique, patented miniature fuel-cell car and hydrogen refueling station. All you need to do
is add water. Only $115.00 plus S/H
Robotics Demystified
by Edwin Wise
YOU DON'T NEED ARTIFICIAL INTELLIGENCE
TO LEARN ROBOTICS!
Now anyone with an
interest in robotics can
gain a deeper under-
standing without for-
mal training, unlimited
time, or a genius IQ. In
Robotics Demystified,
expert robot builder
and author Edwin Wise
provides an effective and totally painless
way to learn about the technologies used
to build robots!
$19.95
Pub Date October 20, 2004
The Day the Earth
Stood Still
An alien (Klaatu) with his
mighty robot (Gort) land
their spacecraft on Cold
War-era Earth just after
the end of World War II.
They bring an important
message to the planet
that Klaatu wishes to tell
to representatives of all
nations. Directed by: Robert Wise
Actors: Michael Rennie, Patricia Neal
Run Time: 92 minutes $14.95
Forbidden Planet:
50th Anniversary Edition
A starship crew goes to
investigate the silence of a
planets colony only to
find two survivors and a
deadly secret that one of
them has. Dr. Morbius and
his daughter Altaira have
somehow survived a
hideous monster which
roams the planet.
Directed by: Fred M. Wilcox
Run Time: 98 minutes $26.95
Blade Runner
(The Directors Cut) (1982)
In a cyberpunk vision of
the future, man has devel-
oped the technology to
create replicants, human
clones used to serve in the
colonies outside Earth but
with fixed lifespans. In Los
Angeles, 2019, Deckard is a
Blade Runner, a cop who
specializes in terminating replicants.
Directed by: Ridley Scott
Rated: R Violence $19.95
CLASSIC SCIENCE FICTION DVDS!
With this kit, you now can see and
feel the future of energy generation
in your own hands!
Linux Robotics
by D. Jay Newman
If you want your robot
to have more brains than
microcontrollers can
deliver if you want
a truly intelligent,
high-capability robot
everything you need
is right here. Linux
Robotics gives you step-
by-step directions for Zeppo, a
super-smart, single-board-powered robot
that can be built by any hobbyist. You also
get complete instructions for incorporating
Linux single boards into your own unique
robotic designs. No programming experi-
ence is required. This book includes access
to all the downloadable programs you need,
plus complete training in doing original
programming. $34.95 Sale Price $ 29.95
Or order online www.servomagazine.com
SERVO 02.2008 75
W
ARNING: This Appetizer con-
tains opinions which may not
taste good to the robotics community!
Consume with caution!
Its New Years Eve, Im 50 years
old, and instead of partying until Im
stupid, Im home writing a column for
SERVO. Its amazing how a few years
change things! Three decades ago,
I thought New Years was the best
party opportunity of the year;
now its a quiet night to write. Im
reminded of the changes time brings
because the soundtrack to this
evenings writing is my wifes
Christmas present: an iRobot Roomba
cleaning robot clattering and
bumping around. Its probably just a
coincidence (but who knows what
lurks deep in the brain) that I just
finished my umpteenth re-read of the
first sci-fi book I ever read, Robert
Heinleins The Door Into Summer.
In that book (published in 1957, the
year I was born), Heinlein clearly
predicted the automated floor clean-
ing robot in his Hired Girl machine.
It took 50 years all the way into the
third millennium before his dream
was realized as a commercial success.
In the same book, he predicted
CAD via his Drafting Dan, although
he missed a bit in that his version was
mechanical in nature, rather than
computerized. Still, CAD became a
reality just one short decade after he
predicted it, so lets score him with a
close enough.
This train of thought led me
to consider where the world is,
robot-wise, in 2008.
Im a solid member of the combat
side of robotics, but with interest in
other competitive aspects of the sport,
and in the overall implementation of
the technology in our society. I have
very mixed feelings about the state
of things, robotically speaking, in
todays world.
First, the down side. The general
feeling in the combat community is
that things continue to decline. There
were about 25 events held in 2007,
down from 35 in 2006, and with
generally smaller attendance. At the
larger events, fewer are travelling long
distances to participate, and there is
a definite downturn in both new
builders, and in large class bots.
In the consumer market, as well
as industrial, I dont see the kind of
rapid evolution in robotics technology
as there is in computers, cell phones,
and music playback. This may be a
natural outcome of the mechanical
nature vs. electronic, the lack of a
driving economic force, or as I
suspect a symptom of the gradual
shift of our culture from an outdoor,
farm based, mechanical nature to an
indoor, information based, electronic
nature. The closing down of so many
high school auto, machine, and wood
shop classes, being replaced with
web, digital, and computer aided
design classes, seems to back up
this assessment.
In hobby and educational robotics,
Im disappointed in the overall lack of
leadership and integration. There are
many complementary programs and
types of competitions, but they
dont seem to cooperate, or even
communicate, well. I, for example, had
no idea Science Olympiad even had a
robotics element until I was called by a
parent for help. (Im known as that
robot guy at work.) The lack of
outward cooperation between BotsIQ
and RFL member groups is another
concern of mine.
Now, more within my comfort
zone, the up side. FIRST has done a
tremendous job of building support at
that key level the state education
department. BotsIQ has a solid
membership base, and is holding
regional tourneys in addition to the
nationals. RoboGames 2007 pulled in
a whopping 559 registered bots, from
all around the world, in the widest
possible range of categories. In
combat, the 25 events in 2007 in
North America represent one every
two weeks. A few RFL member
organizations and individuals have
programs working in their local
schools, and some participants in
BotsIQ also cross over to local,
regional, or national combat events.
The Door
Into Spring
by Kevin Berry
76 SERVO 02.2008
Appetizer.qxd 1/10/2008 11:11 AM Page 76
The war in Iraq and Afghanistan
has spurred tremendous develop-
ment in robotics in that toughest of
all environments: desert and urban
warfare. UAVs, remotely operated
ground machines, and fusion of sen-
sor data with manned platforms
have pushed the kind of evolution I
mentioned above. Another driving
factor in robotics is the sudden,
horrific need for prosthetic
replacements for the loss of limbs in
our combat troops. This is an area of
robotic technology that has, in my
opinion, been pushed over the
tipping point and is about to
become consumer driven.
The military is also, indirectly,
pushing robotics through programs
like the DARPA Grand Challenge.
A similar activity from the space
community the privately funded X
prizes is also moving things along
nicely in robotic developments.
Another hot topic climate
research is forcing developments in
robotics, in the underwater, Arctic,
and aerial environments. Universities,
fed by government grants and
spurred by various competitions, are
developing many lines of technology
that directly, or indirectly, relate to
robotics development.
Now that Ive spent 800 words
assessing, the reader might ask,
so, whaddya gonna do about it?
A valid question, and one for
which I only have partial answers.
There has been enough written
about the state of science and
engineering education in the USA
to fill a large library (7,410,000 hits
on Google, btw). So I think Ill
confine myself to some more
specific comments.
First, obviously, the higher the
level the initiative is being
discussed, the less any individual
can do about it. To put together the
kind of support Dean Kamen has
(with his landing of the Rhode
Island Department of Educations
commitment to include every high
school in the state in the VEX
competition) is beyond the
resources of most of us.
However, focused lobbying by
robotics clubs to the members of
their state legislatures Education
committees is very practical and
achievable. Leadership, in the form of
a grass roots national organizing
effort, would require a few far-
sighted individuals to put together
a boilerplate package including a
summary of existing programs,
contacts, organizations, and an
implementation plan. This could
then be used by state level ad hoc
groups, composed of members from
different clubs and educational
efforts, to lobby the state committee
membership. Demonstrations of
existing (albeit probably fragmented)
robotics programs in their districts
would naturally follow.
Secondly, and more achievable by
the kind of dynamic individuals found
in the robotics community, is cross-
communications and cooperation
between programs and clubs at the
national, regional, and local level.
There is, in my humble (ahem!)
opinion, absolutely no excuse for
elitism or snobbery among us. For
FIRST folks to look crossways at
combat folks, or R/C folks to snub
autonomous competitions, is just
plain dumb. We need more, region-
al/state/local versions of RoboGames,
where all types of competitors come
together in robotic fellowship.
Third, and totally do-able by
SERVO subscribers, is individual
involvement. There are, very roughly,
105,000 K-12 schools in the US. If
every American reader of this
magazine volunteered to do an
in-school or after-school robotics
program within their individual areas
of interest, we could heavily impact
the state of robotic education in this
country. Realistically, given the 80/20
rule and irregular distribution of
readers, we might be able to
support 10,000 or so schools, either
expanding their existing efforts or
bringing new schools into the fold.
At five kids per school, thats a huge
kernel of future roboticists to push
forward into college!
In conclusion (ah, those
treasured words), Im pretty
optimistic that the next five decades
will see great leaps in the integration
of robotics in the home, workplace,
school, and general society. While
much of the impetus will come from
market forces and human need, its
a proven fact that a groundswell
of public support can also move
our society in a direction, even
counter to the market. This field sits
where the computer sat in 1990, just
coming into the springtime of the
growth curve. How fast we move
from spring to summer is the
variable. Since I probably wont see
the year 2057, Id sure like yalls help
in making it happen quicker, maybe
say, 2017? SV
SERVO 02.2008 77
Appetizer.qxd 1/10/2008 11:11 AM Page 77
78 SERVO 02.2008
I
ve written about personal robots for
years in this column. Ive discussed
some of the early machines available
back in the late 70s and early 80s
such as the Heath Hero series, the
Androbots, and the RB5X, among oth-
ers. These machines certainly differed
from their industrial cousins that toiled
away in modern factories. They also
differ from service robots that include
ROVs (remotely operated vehicles),
UAVs (unmanned aerial vehicles) that I
wrote about last month, and AUVs
(autonomous unmanned vehicles) that
go to places that are too dangerous for
people to visit or too expensive to send
them there such as the battlefield,
undersea, or deep space.
This latter category can also
include the DARPA Grand Challenge
vehicles built from cars and trucks,
teleoperated surgical and nuclear hot
cell robots, window washing and
sewer inspection robots, and even
competitive combat robots all robots
that are normally not used in the
presence of humans. (And then there
are the robots that move about and
execute tasks amongst humans, like
the daVinci surgical robot.)
Thats a lot of categories! Rather
than discuss improvements throughout
the history of personal robots, Im
going to go over some of the design
obstacles and stumbling blocks that
Ive seen over the years.
Robots at Our Service
This very special category of service
robots has the ones that most of us who
are reading SERVO Magazine would
really like to have in our homes. We look
favorably at Rodney Brooks Cog that he
and his team worked on at MIT several
years ago; a personal robot that was
designed to have a personality tied to
facial expressions. Figure 1 (taken by
Sam Ogden) shows COG in a pose
simulating deep thought. David Hanson
has taken that a bit further with
extremely realistic robotic heads with
very sophisticated facial expression
mechanisms. Feature films such as
I-Robot have made us feel as if truly per-
sonal robots are just around the corner.
We want one in our homes NOW
and experimenters, various manufactur-
ers, and researchers are rapidly moving
towards the goal of developing a true
personal robot for all of us. Whether
we are beginners experimenting with
hobby servo-driven robot kits or are a
bit further along in robot design, for
most of us, the final goal is to build a
true personal or domestic robot that
can actually accomplish tasks for us.
The Ultimate Robot
Robot experimenters, designers,
and manufacturers always like to call
their latest creation the ultimate
robot. They feel that their pride and
joy has finally reached that pinnacle of
development; it will deliver any and
all functions that the prospective
buyer may want. I must admit that I
once called a robot that I designed
Ultima, implying that it was indeed, the
ultimate robot.
Featured in an April 1984 article
that I wrote for Popular Mechanics, I
felt that the ultimate robot should have
very powerful arms so I used two
off-road style winches to drive the
shoulder joints. These suckers drew
over 100 amps each and the robots
ultimate claim to fame was eating
batteries for dinner, or at least rapidly
discharging them. Calling my creation
the ultimate robot would make about
as much sense as an observer of the
Wright brothers first plane flight (of a
minute or so at walking speeds)
stating: The ultimate aeroplane will be
able to stay in the air over an hour,
carry two persons, and travel 50 miles.
It all depends on your point of view
and the current state of the art.
Revenge of the
Nerds Robots
Everyone wants the ultimate
a
n
d
PERSONAL ROBOTS:
From Science Fiction to Reality
b y T o m C a r r o l l
FIGURE 1. COG.
Then&Now.qxd 1/10/2008 11:37 AM Page 78
robot. Years ago, when Twentieth
Century Fox studio management had
read about my robots, they asked me if
I could build a robot for their upcoming
movie, Revenge of the Nerds. I told
them that the design and building of a
robot prop would be no problem. I had
the shop, machine tools, all the parts,
and friends who could help out in a
pinch (as I had a regular job at
Rockwell). Then, the producer began to
outline his idea of the ultimate robot.
Articulated hands, numerous
special effects functions, and even legs
to walk with seemed easily obtainable
by the producer. After all, he had seen
so many robots in movies and only
vaguely realized that most of these
were people in robot suits.
He really had no clue about the
complexity of what he desired. I
quickly assured him that walking
robots were out of the question but he
still wanted arms that moved at the
shoulder and elbow, and hands that
could grip things.
Figure 2 shows one of the
completed Revenge robots. A figure 8
cable from the shoulder to the elbow
gave motion to the elbow from the
single shoulder motor; the R/C control
system from Vantec controlled all the
motors and special effects systems.
I delivered four robots: two full-
blown action props with all systems
and two partial props for non-moving,
static shots in less than three weeks.
These teleoperated robots survived the
shooting schedule, but I have no idea
what happened to them afterwards.
I later sold an acquaintance an old
robot that I called Squirt that I made
from a blue plastic chemical drum that
was featured in the third Revenge of
the Nerds film. That was over two
decades ago and the only feedback
that existed on the movie set was to
and from the special effects prop
guys off scene with hand-held R/C
transmitters in their hands.
Building props for movies was very
interesting back then, but it is now
easier to build things for films with
computer graphics and simulations.
Were these early builds the ultimate or
even cutting edge personal robots? No
way. These were just R/C props with
absolutely no autonomy.
Personal Robot
Autonomy for
Experimenters
Todays robot experimenter has far
better software, feedback encoders,
sensors, motor systems, and micro-
controllers at his or her disposal and
demands better performance for the dol-
lars spent. They dont want teleoperated
machines. They want a machine that
senses its own environment; a machine
that will interact with that environment
and the people within the environment
an intelligent robot. Theyve seen
Hondas Asimo and all the humanoid
robots that are available and desire a
robot that can walk or roll amongst us,
serving and conversing with us.
A few of the non-technical have
seen movies such as I-Robot where
robots run amok and are a bit wary of
the offerings of the future. What can
we experimenters do to advance the
robotic state of the art, and, at the
same time, convince the general public
that robots are quite harmless and are
here to stay? Do we dare tell our non-
technical friends that Asimo can only
operate a few minutes at a time with
his heavy battery backpack and with
Honda personnel behind the scenes,
nervously worrying that their $200K
creation may tip over at any moment?
The Robot Servant
From the many topics and threads
that Ive followed on the websites of
different robot clubs, a servant robot
always seems to pop up as that most
desirable robot. Kids want a robot to
do their household chores and clean
their rooms. When we get out into the
world on our own, besides mowing the
lawn and vacuuming the floors, bring
me a Pepsi (or beer) and some chips
during the big game seems to be an
oft asked question of our robot-to-be.
We pretty much know that
cleaning a room is virtually impossible
with todays technology as there are
so many variables in a constantly
changing environment. Clothes on the
floor, toys and sports equipment
strewn about, constantly moving
chairs, hard-to-reach shelves with clut-
ter, dust everywhere, and dried pizza
sauce on a table top are just a few of
the difficult obstacles. A typical room
needing cleaning is a three-dimension-
al target requiring extreme dexterity
and sophisticated vision systems and
software that has yet to be written. It
is so much easier to use a Roomba to
do the basic floor cleaning and leave
the more complex chores to a human.
Maybe you secretly desire a robot
dog a mechanical hound like the one
that the firemen kept in their firehouse
in Ray Bradburys classic Fahrenheit
451. Of course, this hound did not sit
on the speeding fire truck, ears a
flappin; it was more adept at injecting
captured rats with morphine through a
four-inch needle. Possibly an old Aibo is
more your speed, or a robot that can
fetch your slippers and daily paper.
The Fetching Task
The fetching robot asked to bring a
Pepsi and chips to its owner is a bit
more obtainable by todays robot
designer. Certainly any part of a typical
home is far more complex than the
structured environment of a factory
where the industrial robot is in a fixed
location and pretty much knows where
its work pieces are. For a fetching task,
the robot usually has to traverse a single
two-dimensional floor. The problem
may be not only what is on the floor,
but what is now on that floor that was-
FIGURE 2. Revenge of the Nerds robot.
SERVO 02.2008 79
80 SERVO 02.2008
nt there the last time the robot traveled
it. There will always be obstacles, walls,
doorways and an optimum path to
follow, but, there may be unknowns
that the programmer can not anticipate.
Retrieving a beverage may bring
forth the following thought processes
in the robots programming:
Bring means I am to initiate a
series of processes and set forth a
series of motions to leave my present
location and go to another location,
set forth another series of motions to
grasp a requested object and return
the object, not to where I started, but
to the location of the requestor.
Me refers to the requestor, so I
have to know his/her location before I
start my retrieval procedure so I can
return the requested object to that
location. The robot scans the room,
maybe using a directional microphone
to locate a voice, possibly a PIR sensor
to detect a 98.6 human, judge the
distance by an ultrasonic transducer,
look for obstacles, and determine the
location of the requestor.
Pepsi is a food item, and my mem-
ory tells me that this particular food
item is in the refrigerator. A refrigerator
is a special storage device. It has a door
that swings outward to the left in a
30 radius arc. It also requires a bit of
a jerking motion to unseal the door,
and then I must swing my arm inward
to bring the door open to 90 degrees.
The Pepsi is on a shelf 27 above
the floor and its bar code is facing
outward (previously arranged by the
owner) so I can scan the shelf with the
bar code scanner built into my hand
until I locate a Pepsi, then move
my gripper forward until the contact
sensor detects the fingers of the
gripper completely surrounding the
Pepsi. I then close my gripper with a
slight force (enough to hold a wet 12
oz. can) and retract my arm and
gripper from the refrigerator.
After reversing the above proce-
dures of closing the door, turning
around 180, and heading back on
its previous path, the robot must
now change its path and return to the
location of the requestor. Here is your
Pepsi, Master!
Is it all this simple? Not on your
life! Of course, robots dont process
data like this. This is a human thought
process. However, if we were behind
the controls of this robot, we would
have to go through these thinking
steps and far more to accomplish
such a simple task.
CPUs do it differently. There are so
many routines and subroutines to
consider, sensors to continually read
and update the program, changes in
the pathway, and what/ifs to consider.
What if the kitchen door is closed and
opening it is not part of the robots
programming? What if there are no
more Pepsis in the fridge? What if the
requestors voice has not been
programmed into the robots speaker-
dependent speech recognition system?
What if his dog layed down in the
intended return pathway after the
robot went to the fridge? What if the
requestor asked for a Mountain Dew or
a Diet Pepsi instead?
When do you finally stop adding to
your programming? Will the requestor
have to say, No, Robbie, thats the
dog youre detecting with you PIR
sensor; Im over here. When can the
robot solve situations that the
programmer never anticipated and
added to the program? As far as I
know, there are few robots that truly
solve problems; they only run the
programs that we place into them.
How about the robots navigation?
GPS is usually not applied to robot
navigation in a small home environment
with poor signal reception, so bit map-
ping, virtual mapping, or other methods
are sometimes used. Experimenters
sometimes apply bar codes or active IR
beacons to strategic locations within the
home to identify things like doorways.
RFID tags are becoming popular
and can be used to identify common
household objects. I once used an RF
energized wire applied under new
carpeting for a robot to follow. An
invisible UV or IR dye can be applied to
carpeting for the robots onboard
IR/UV lamp to activate and follow
with appropriate sensors. Ultrasonic or
shaped-beam IR/visible LEDs can detect
walls and hallways. Waypoints can be
established around fixed objects to
simplify the robots navigation.
Limitations of the
Ultimate Personal
Robot
Do experimenters really care to
design the ultimate robot? Isnt a series
of incremental improvements over a
period of time really more important
for the learning process? Do we need a
robot that walks about and emits
canned words that people understand
or do we need a smart robot that can
respond to human conversation?
I feel that the appearance of
intelligence seems to be the goal of
many robot builders; maybe an AI
program that is close to passing the
Turing Test of intelligence. We realize
that human-sized walkers like Asimo are
beyond the average experimenter, but
the integration of a good AI program
with top notch speech recognition may
bring our carefully-crafted creations clos-
er to what we call the ultimate status.
In the end, well end up with
something similar to an R2D2 instead
of a C-3PO, but the robot will have an
extensive language instead of a series
of unintelligible beeps.
Robot Artificial
Intelligence
The AI part may have to undergo a
bit of change to understand the many
quirks of the English language, or any
of the worlds languages, for that
matter. So many words have different
meanings, and basic grammar and
context are difficult, if not impossible
to program into todays speech
recognition systems. Throw in colloqui-
al expressions, slang terms, and incom-
plete sentences and you have a really
confused robot. Add in echoes from
around a room, variances in sound
levels and youll need a pretty good AVC,
preamp and band pass filter system.
Vision systems available to the
experimenter such as the series of
CMUCAMs, AVRcam, and the LEGO-
CAM provide quite a bit of visual intel-
ligence for robots, but they still need
refinement for true personal robots
operating in complex environments.
The truly autonomous robot will
need to have a suite composed of
numerous sensors. The software
involved will not only have to make
sense of each one, but be able to
distinguish which of the many sensor
inputs are applicable to the single line
of code that the microcontroller/micro-
processor is reading at any moment.
The experimenter will have to take
each sensor whether a motion-
encoding device, a navigation/position
sensor, environmental sensor, vision or
speech system, or any other type and
produce the code to make sense of
each sensors output and how it affects
the robots particular situation.
The Ultimate Robot
Control
Control? youre possibly asking
yourself. I thought we were looking
at the ultimate in autonomy in robots
here. Well, lets face it, even with
autonomy, robots will need directions
from us. Thats control. Perhaps
thought control may be a better way to
go instead of verbal commands.
Texas Instruments recently
demonstrated a thought-controlled
wheelchair in India. The Ambient
Corporation chair used a system called
Audeo and one of TIs MSP430
microcontrollers. To move the chair
in one of four directions, the user
needed only thought patterns and
slight movements of his Adams apple.
Of course its not true thought
control and it leaves out female use of
the chair, but maybe Cyberkinetics
Neurotechnology Systems implantable
BrainGate microelectrode array is a
better way as that has been used to
control a television set and a prosthetic
hand. Hmm. Brain surgery? Lack of an
Adams apple? Maybe we have a ways
to go to develop ultimate robot control.
Is There an
Ultimate Robot?
In the preceding paragraphs, Ive
used the word ultimate many times to
describe all types of personal robots. In
doing so, Ive pretty much convinced
myself that any robot having such a
title will not hold that moniker
very long. Progress continues to march
forward and Japan will always be in the
forefront of personal robot develop-
ment. Waseda University has always
ranked first or second in the world
with advanced robot research and
development, with MIT, CMU, or
Stanford occasionally bumping them
from first place.
Wasedas Sugano Laboratory
recently unveiled Twendy-One (Figure
3) after seven years of development
and several hundred million Yen
expended. The robot has 241 pressure
sensors on each of its amazing silicone-
covered hands and can assist a person
out of bed and serve them meals, yet is
hampered by a very short 15 minute
battery life a problem that hampers
so many advanced robots.
Notice the cable behind the robot.
At about five feet tall, it weighs a hefty
245 pounds. It is hoped that Twendy-
One can be available in the homes of
the elderly by 2015, at a cost of about
$200,000.
A key feature in Twendy-Ones
design is its passivity very important
in working around fragile humans.
Compliance in the arms and body parts
in conjunction with numerous touch
and force sensors allows the robot to
react to collisions in order to prevent
injuries in a human-symbiotic relation-
ship. I, personally, hope to develop a
personal robot assistant under $20,000
to allow seniors independence in daily
living a tall order.
Not to be outdone,
Stanford University has
developed STAIR 1.0, a
personal robot mounted
atop a modified Segway
platform (Figure 4). Note
the two casters fore and
aft to prevent the robot
from toppling to the floor
in the event of a power
failure. A SICK intelligent
sensor and articulated
arm allow the robot to
interact with humans and
perform simple fetching
tasks, all with verbal
commands.
The UMan from the
University of Massachusetts
is a similar robot using the same hand
as STAIR (developed by Barrett
Technology) and similar sensors (Figure
5). The UMan is continually being
trained to sense new and familiar
objects and manipulate them in a
manner useful to serve humans.
What is a Personal
Robot?
Thats a good question! Just what
should the ultimate personal robot be
capable of doing? Should it be capable
FIGURE 3. Twendy-One.
FIGURE 5. UMan. FIGURE 4. Stanfords
STAIR 1.0 robot.
SERVO 02.2008 81
of operating up and down stairs
(considering not all people have
stairs)? Should it be human height or a
bit shorter with a low center of gravity?
How about mobility should it be able
to operate outdoors in rough terrain
and bad weather, as well as in a
cluttered environment inside?
Should it walk on two feet or
more, or just roll along? How about
speech recognition; navigation;
articulated arms with five-fingered
hands; soft, pliable skin or hard exterior?
How about the ability to lift a human in
need or just heft a can of Pepsi?
There are an estimated four
million+ personal or domestic robots in
service today. Though most of these
are probably Roombas, there are
probably four million different ideas
out there for the ideal personal robot.
Hopefully, this will stir your
creative juices to create a robot that
will be the killer app. for home/
domestic personal robots. I am firmly
convinced that the best robot
research is coming from independent
experimenters like SERVO readers,
toiling away in their garages or
basement workshops to improve a
small facet of a robots design.
By sharing your knowledge in
magazines such as SERVO, as well as
on the many robot groups websites,
we will all come up with the ultimate
robot. As always, I really appreciate
your feedback on my articles. SV
Tom Carroll can be reached via email
at TWCarroll@aol.com
CONTACT THE AUTHOR
All Electronics Corp. .........................53, 56
AP Circuits/e-pcb.com ..............................7
AWIT ..........................................................56
Boca Bearings .....................................56, 61
Budget Robotics ......................................77
CipherLinx Technologies .........................56
CrustCrawler ...............................................3
Electronics123 ..........................................53
Futurlec .....................................................56
Hitec ............................................................2
Jameco ......................................................66
Lorax Works ........................................53, 56
Lynxmotion, Inc. .......................................13
Maxbotix ...................................................56
Net Media .................................................83
Online-tech-training.com .........................56
Parallax, Inc. ...............................Back Cover
PCB Pool ..............................................12, 56
Pololu Robotics & Electronics ..........38, 56
Robotis Co. Ltd. ..........................................9
RobotShop, Inc. .................................56, 82
Schmartboard...........................................53
Solarbotics/HVW .....................................19
Technological Arts ...................................56
Vantec .........................................................7
Advertiser Index
82 SERVO 02.2008
backcvr.qxd 1/7/2008 1:43 PM Page 84

Servo 2008 - 02

Uploaded by

Copyright:

Available Formats

Servo 2008 - 02

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Servo 2008 - 02

Uploaded by

Copyright:

Available Formats

V

You might also like