Trajectory Generation

Trajectory Generation
Amirkabir University of Technology

Computer Engineering & Information Technology Department
Introduction
How to compute a trajectory in multidimensional space
which describes the desired motion of a manipulator?
How to move a manipulator from its initial position to

some desired goal position in a smooth manner?
Trajectory
A time history of position, velocity, and
acceleration for each DOF.
The system (not the user) will decide the exact
shape of the path to get to the goal, the duration,
the velocity profile, and other details.
A sequence of desired via points (frames which
specify position and orientation of tool) can be
given.
Path points include all the via points plus the
initial and final points.
General Considerations
Each path point is specified in terms of a
desired position and orientation of {T}
relative to {S}.
Each of these points is converted into a set
of desired joint angles. (Inverse kinematics).
A smooth function is found for each of the n
joints which pass through the via points and
end at the goal point.
The time required for each segment is the
same for each joint.
Why Smooth Motion?

Increased wear on the mechanism
Vibrations
A smooth function is continuous and has a continuous first

(and second) derivatives.
Joint Space Scheme

The path is specified in the joint space.
Easy to compute.
No problems with singularities.
Each path point is usually specified in
terms of desired position and
orientation of tool frame {T}.
Each of these via points is converted
into a set of desired joint angles by
inverse kinematics. Then a smooth
function is fond so that all joints will
reach the via point at the same time.
Several Possible Path Shapes

for a Single Joint
Goal
position
Initial
position
Cubic Polynomials
4 constraints on (t )
(0) 0 , (t f ) f ,
(0) 0, (t ) 0.
: initial and final values

: the function is continuous
in velocity
These 4 constraints can be satisfied by a polynomial

of at least third degree.
(t ) a0 a1t a2t a3t

(t ) a 2a t 3a t 2
2
Cubic Polynomials
0 a0 ,
f a0 a1t f a t a t ,
2
2 f
3
3 f
: combining with constraints
0 a1 ,
0 a1 2a2t f 3a3t 2f .
a0 0 ,
a1 0,
3
a2 2 ( f 0 ),
tf
2
a3 3 ( f 0 ).
tf
The cubic polynomial that

connects any initial joint angle
position with any desired final
position when the joint starts and finishes
at zero velocity.
Example 7.1
A single-link robot with a rotary joint:
Move the joint in a smooth manner
from =15 to =75 in 3 seconds
a
a
a
a
15.0
0.0
20.0
4.44
(t ) 15.0 20.0 t 4.44 t

2
(t ) 40.0t 13.33 t
(t ) 40.0 26.66t
Cubic Polynomials for a Path

With Via Points
Each via point is specified in terms of a desired
position and orientation of {T} relative to {S}.
Each of these via points is converted into a set
of desired joint angles. (Inv. Kinematics).
Consider the problem of computing cubics
which connect the via point values for each
joint together in a smooth way.

With Via Points
(0) 0 , (t f ) f .
: some known velocity

at the via points
0 a0 ,
f a0 a1t f a2t 2f a3t 3f ,
: the four equations
0 a1 ,
describing general cubic
a 2a t 3a t 2 .
f
2 f
3 f

With Via Points
Solving for ai we obtain
a0 0 ,
a ,
1
3
2 1
a2 2 f 0 0 f ,
tf
tf
tf
2
1
a3 3 f 0 2 f 0 .
tf
tf
Connects any initial and final positions with any initial

and final velocities
How to Specify Desired

Velocity at the Via Points?
Several ways:
The user specifies the desired velocity at each
via point in terms of a Cartesian linear and
angular velocity of {T} at that instant.
The system automatically chooses the velocities
at the via points.
The system automatically chooses the velocities
at the via points in such a way as to cause the
acceleration at the via points to be continuous.
A simple heuristic
Tangents to the curve at each via points

If the slope of these lines changes sign at the via point, choose zero
velocity, if the slope of these lines does not change sign,
choose the average of the two slopes as the via velocity.
Higher Order Polynomials

If we wish to be able to specify the
position, velocity, and acceleration at the
beginning and end of a path segment, a
quintic polynomial is required.
(t ) a0 a1t a2t a3t a4t a5t ,
2

(t ) a0 a1t a2t 2 a3t 3 a4t 4 a5t 5 ,
0 a0 ,
:the constraints are given as
f a0 a1t f a2t 2f a3t 3f a4t 4f a5t 5f ,

0 a1 ,
a 2a t 3a t 2 4a t 3 5a t 4 ,
f
2 f
3 f
4 f
0 2a2 ,
f 2a2 6a3t f 12a4t 2f 20a5t 3f .
5 f

The solution for a linear set of equations with six unknowns:
a
a
0
0
2
a3
20
20 0 (
12
2t 3
)t f (
3
0
)t 2
a4
30 0 30 f (
14
16
2t 4
)t f (
12
12 0 (
2t 5

)t f (
)t 2
)t 2
Linear Interpolation
Another choice of path is linear:
Discontinuous velocity increase at the beginning and

discontinuous velocity decrease at the end of the motion,
requiring infinite acceleration.
Linear Function With Parabolic

Blends
Constant
acceleration
The linear function and the two parabolic functions are splined
together so that the entire path is continuous in position and
velocity.

Blends
b
The parabolic blends have the same duration,

are symmetric about the halfway point in time,
and the halfway point in position.

Blends
The velocity at the end of the
t h b ,
b
t h tb
blend region must equal the

velocity of the linear section.
1 2
b 0 tb .
:One choice is.
2
t 2t h
tb2 ttb ( f 0 ) 0
2t 2 4( f 0 )
t
tb
2
2
4( f 0 ) . : the constraints
on acceleration
t2
When equality occurs, the linear portion has shrunk to zero length.
Example 7.3
Two possible choice
of linear path with
parabolic blends for
the path described in
example 7.1
Linear Function With

Parabolic Blends for a Path
With Via Points
Linear function connects the via points and parabolic

blend regions are around each via point

With
Via Points
For 3 neighboring
path points: j,k,l
: duration of blend region at K
jk
: duration of linear region between j,k
djk
: overall duration of segment connecting j,k
jk
: velocity during linear portion

: acceleration during the blend

With
Points
We canVia
compute
the blend time:
k j
jk
,
tdjk
k SGN (kl jk ) | k |,
kl jk
tk
,
t jk tdjk
1
1
t j tk .
2
2

With Via Points
The first and the last segments might be

handled slightly different since an entire
blend region must be counted in the
segments time duration:
This can be solved for blend

time at initial point as:
1
t d 12 2 t1
t
1
SGN ( 2 1)
t t
1
d 12
2(
d 12
1
t d 12 2 t1
1
t12 t d 12 t1 2 t 2
12
The Path Generator

The results of computations
constitute a plan for the trajectory.
At execution time the path
numbers
generator will use these
to compute , ,
Use of Pseudo Via Points to

Create a Through Point
Via points are not actually reached unless manipulator comes
to stop. If we wish exactly passing through a point without
stopping, the system may replace via point with 2 pseudo via
points.
A path point through which we force the manipulator to pass exactly!
Cartesian Space Schemes

We can specify the spatial shape of the path
between path points. (Straight, circular,
sinusoidal, ).
Each path point is specified in terms of position
and orientation of tool frame relative to station
frame.
This scheme is more computationally expensive
because inverse kinematics must be solved at
the path update rate.
Cartesian Straight Line

Motion
A spatial path which causes the tip
of the tool to move through space
in a straight line can be defined By:
Choosing many via points close
together on a line.
or
Using Cartesian motion functions
to specify a path.
Cartesian Straight Line

Motion
We may use a spline of linear
function with parabolic blends for
Cartesian motion. However as a
rotation matrix must be composed
of orthonormal columns and this
condition would not be guaranteed
if constructed by linear
interpolation.
Angle-axis Representation
Angle-axis representation can be used to specify an
orientation with 3 numbers. ( for example Euler angles)
If we combine this representation with Cartesian position we
will have a 6x1 matrix:
P
S
K
AORG
A
Now we can describe spline functions which smoothly move

these 6 quantities from path point to path point as function of
time
Angle-axis representation
Note:
Angle-axis representation is not unique
Constraints:
The blend time for each degree of freedom must be the same
to ensure that the motion of all dof will be straight line in
space.
Hence we specify blend time and then compute the
acceleration.
Geometric Problems With

Cartesian Paths
Cartesian path are prone to various
problems related to work space and
singularities.
Intermediate points unreachable

High joint rate near singularity
Start and goal reachable in different
solutions
Cartesian Path Problem

Type I
Intermediate points unreachable
Moving from A to B
would be no
problem in joint
space, but the
straight path
contains unreachable points in
Cartesian motion.
Workspace
violation

Type II
High joint rate near singularity
Near singular configuration

Type III
Start and goal reachable in different solutions
Joint limits
Next Course:
Robot Programming
Amirkabir University of Technology

Computer Engineering & Information Technology Department

Trajectory Generation

Uploaded by

Copyright:

Available Formats

Trajectory Generation

Uploaded by

Document Information

Original Description:

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

Trajectory Generation

Uploaded by

Copyright:

Available Formats

Trajectory Generation

Amirkabir University of Technology

How to move a manipulator from its initial position to

Why Smooth Motion?

A smooth function is continuous and has a continuous first

Joint Space Scheme

Several Possible Path Shapes

: initial and final values

These 4 constraints can be satisfied by a polynomial

(t ) a0 a1t a2t a3t

: combining with constraints

The cubic polynomial that

(t ) 15.0 20.0 t 4.44 t

Cubic Polynomials for a Path

Cubic Polynomials for a Path

: some known velocity

Cubic Polynomials for a Path

Connects any initial and final positions with any initial

How to Specify Desired

Tangents to the curve at each via points

Higher Order Polynomials

Higher Order Polynomials

:the constraints are given as

f a0 a1t f a2t 2f a3t 3f a4t 4f a5t 5f ,

Higher Order Polynomials

Discontinuous velocity increase at the beginning and

Linear Function With Parabolic

Linear Function With Parabolic

The parabolic blends have the same duration,

Linear Function With Parabolic

blend region must equal the

Linear Function With

Linear function connects the via points and parabolic

Linear Function With

: duration of blend region at K

: duration of linear region between j,k

: overall duration of segment connecting j,k

: velocity during linear portion

Linear Function With

Linear Function With

The first and the last segments might be

This can be solved for blend

The Path Generator

Use of Pseudo Via Points to

A path point through which we force the manipulator to pass exactly!

Cartesian Space Schemes

Cartesian Straight Line

Cartesian Straight Line

Now we can describe spline functions which smoothly move

Geometric Problems With

Intermediate points unreachable

Cartesian Path Problem

Cartesian Path Problem

Near singular configuration

Cartesian Path Problem

Amirkabir University of Technology

You might also like