Faculty of Engineering – Cairo University

Mechanical Design and Production Engineering

Two-Semester System – 3rd Year
MDP 307: Modelling and Simulation of Dynamic
Systems and Robots– Fall 2019

Report On
(Reconfiguration Modular Robots)

Submitted to
A.P: Antoine S. Dimitri
A.P: Yasser H. Anis

Prepared by Group F
Mahmoud Ahmed EL-Sayed
Mahmoud Mohamed Tawhid
Anton Tamer Shafique
Ayman Hisham Mahmoud

December 18th, 2019

 Table of Contents

List of Figures ................................................................................................................................. 3

Abstract ........................................................................................................................................... 4
Introduction ..................................................................................................................................... 5
History............................................................................................................................................. 6
What are modular robots? ............................................................................................................... 6
Types of self-reconfiguring modular robots: .................................................................................. 7
Lattice architectures ........................................................................................................................ 7
Definition: ................................................................................................................................... 7
General mechanism:.................................................................................................................... 7
Example robot: ............................................................................................................................ 7
M-Blocks 2.0: Self-assembling Modular Robots: .................................................................. 7
Chain/tree architectures .................................................................................................................. 9
Definition: ................................................................................................................................... 9
General mechanism:.................................................................................................................... 9
Example robot: ............................................................................................................................ 9
PolyBot: .................................................................................................................................. 9
Hybrid architectures ...................................................................................................................... 10
Definition: ................................................................................................................................. 10
Example robots: ........................................................................................................................ 10
Molecubes ............................................................................................................................. 10
SuperBot ............................................................................................................................... 11
Mobile configuration change (MCC)............................................................................................ 12
Definition: ................................................................................................................................. 12
General mechanism:.................................................................................................................. 12
Example robots: ........................................................................................................................ 12
MilliBots ............................................................................................................................... 12
SMORES-EP......................................................................................................................... 13
Future directions ........................................................................................................................... 14
1-Docking and coupling ............................................................................................................ 14
2-Autonomy .............................................................................................................................. 14
3-Field testing ........................................................................................................................... 14
4-Modular manipulation ........................................................................................................... 15
Availability and price .................................................................................................................... 15
Conclusion .................................................................................................................................... 16

Page | 2
 List of Figures

Figure 1: block movement over each other showing the role of magnets on the movement ------- 7
Figure 2: disassembled block module ---------------------------------------------------------------------- 8
Figure 3: components and sensors of block module ----------------------------------------------------- 8
Figure 4: example of PolyBot configuration -------------------------------------------------------------- 9
Figure 5: modules of Molecubes --------------------------------------------------------------------------- 10
Figure 6: FESTO robotic arm ------------------------------------------------------------------------------ 11
Figure 7: example of SuperBot configuration ------------------------------------------------------------ 11
Figure 8: module of MilliBots ------------------------------------------------------------------------------ 12
Figure 9: configuration of MilliBots climbing downstairs --------------------------------------------- 12
Figure 10: several configurations of SMORES-EP ----------------------------------------------------- 13
Figure 11: the motion of SMORES-EP module --------------------------------------------------------- 13
Figure 12: cubelets group set ------------------------------------------------------------------------------- 15

Page | 3
 Abstract

Not very long ago, the idea of having robots in our everyday lives was merely a dream.
Nowadays, robots are invading our lives vastly. Not only have robots been playing a major role in
industry, but even the simplest tasks in our lives, like sweeping the floor. Robots also have various
types now; each has a certain task or a specific way of working. “Modular robots”, or “Self-
reconfiguring robots”, is a term given to a unique type of robots that manages to change its shape
to perform different tasks, giving the robot almost no limitation at all. In this report, we are diving
deeper into the modular robots, going through the origin of the idea itself, the different types of
the modular robot, how it interacts with the environment or the user, and most importantly, what
are the expectations for such a promising robot. Since building a robot that can change its shape
depending on the situation it is facing is a complicated process, there haven’t been many products
in the market. Nevertheless, many scientists and engineers believe in the potentials of the modular
robot, and that –if transpired– will be a solution to most of our unsolvable problems.

Page | 4
 Introduction

Various types of robots are being used around us every day now; whether being used to
explore other planets, heavy industry and mass production, having your car washed, or even a
getting your coffee ready right when you get up. But not all robots are designed the same way, or
manufactured to get the same tasks done, each robot has a specific design to accomplish a certain
task. But how about a robot that can change its own design and shape itself to adapt different
situation? Modular robots are designed for that specific reason. For example, some modular robots
can change its motion method from using wheels to using a set of legs to overcome unsteady
surfaces. The modular robots are divided into other types, depending on the way they’re
programmed, the method they build themselves, or the types of motion they can afford. The
applications for such an extraordinary invention can vary from building games used to entertain
children and nurture their creativity, to rescuing people from places than can’t be reached, to
exploring the space (NASA’s Superbot). On the other hand, the mechanisms used to connect its
building blocks together can sometimes be extremely difficult to design, which creates a huge
problem facing the modular robot’s production, not to mention the complexity of its programming,
or the great cost needed to build a single module of the whole robot. Many obstacles face the idea
of making the modular robot commonly used, yet, its potentials are very promising, and is expected
to be used in various fields soon.

Page | 5
 History

What are modular robots?

“Modular robots” is a term given to one of the newest family of robots, as it is composed
of relatively small building unites called “modules”. Modules, which could be considered as mini
robots, are attached together using various mechanisms. The idea behind these mechanisms is to
allow the robot to reconfigure its shape, enabling it to achieve different tasks depending on the
shape it forms and how these modules are put together.

The beginning of Modular Robots (MRs) can be traced back to 1950s, when the definition
of the universal automata was proposed by John von Neumann as a basic framework of
reconfigurable robots. This made Neumann to be considered as the Godfather of the development
of reconfigurable Modular robots. Although he didn’t develop the physical prototypes himself, his
theory has a far-reaching influence on this development. Then Lionel Penrose proposed the self-
replication model, that can form various structures during the random shaking technique.
Afterward, this area of research was almost forgotten for about two decades. In the late 1970s,
with the development of microelectronics and its wide applications, the area of researching
reconfigurable systems was started again.
In the late 1980s, Toshio Fukuda et al. proposed a new reconfigurable robotic system called
‘‘Dynamically Reconfigurable Robotic System (DRRS).’’ The DRRS consists of a number of
modules can reconfigure their own structure depending on a certain given task. Depending on the
basis of DRRS, Kawauchi and Fukuda carried out a great work on a robot system called cellular
robotic system (CEBOT). Their work was to develop four versions of CEBOT prototypes. This
great achievement in MRs opened the way for other researchers to work on and design various
types of MR prototypes.
In the early 1990s, researchers such as Greg Chirikjian, Mark Yim, Joseph Michael, and
Satoshi Murata worked for further development in lattice reconfiguration systems and chain-based
system. Their development was on designing, building modules and developing codes to program
them. Others such as Daniela Rus and Wei-min Shen worked on hardware to develop it. Then they
started working on provable distributed algorithms to control large numbers of modules.
More recently, Hod Lipson and Eric Klavins pursued new efforts in stochastic self-
assembly. And at Carnegie Mellon University there was a large effort headed by Seth Goldstein
and Todd Mowry who started searching at issues in developing millions of modules.
Recently, researchers have worked on many innovative approaches to design their MRs,
that has promoted the study of MRs to become one of the most prevailing research fields.

Page | 6
 Types of self-reconfiguring modular robots:
Even though modular robots are considered are considered a single type among several
types of robots, modular robots themselves are categorized into different models and architectures.
Modular robots are usually classified into:
• Lattice architectures.
• Chain/tree architectures.
• Hybrid architectures.
• Mobile configuration change (MCC).
• Stochastic reconfiguration.

Lattice architectures
Definition:
Lattice architectures utilize 3D patterns to impose rules for arranging and connecting.
There’s a drawback to this sort of architecture, the motion of modules is limited.

General mechanism:
In the lattice architecture, a certain number of modular robots move in a way they can
rearrange themselves with respect to each other on the grid (lattice). This process could be done
even in the three-dimensional space. For more simplicity, each module should be always
surrounded by a known number of modules during the process. Each module has a cubic shape
like I-Cube and EM-Cube module, this increases the contact area between modules, and by using
its docking interfaces more rigidity can be ensured. motion of individual components on the grid
easier.[1]

Example robot:
M-Blocks 2.0: Self-assembling Modular Robots:

Self-assembling modular robots can change their

geometry according to every requested task. Unlike Rigid
robots which are designed for one task the self-assembling
modular robots obtain -by sensing or the manipulation
navigational- the needed tasks.

Figure 1: block movement over each other showing the role

of magnets on the movement

Page | 7
Components and mechanism:
The process of self-rearrangement that modular robots
make is done by one actuator because inside each
module there is a spinning mass which completely spins
inside the module providing the motion to the body of
the module however the position of the module, this
mass provide variety of motion using just one actuator.
This inertial actuator also provides the ability of jumping
which it’s considered as a unique feature that very few
robots have but on the other hand there is a big challenge
to optimize module for control stability due to applying
a very high amount of energy with respect to a very short
amount of time. This process takes place by sending
codes written on a computer to the modules by a wireless
link, on the other side the module sets its algorithm so
that we called it “complete autonomous” Figure 2: disassembled block module

The module has small volume with cubic shape which

enough to contain:
1. Brushless motor controller
2. Flywheel
3. Braking mechanism electronics
4. Radio
5. Battery
6. Eight base magnets
7. Edge magnets
as illustrated in (Fig. 2) and (Fig.3)

The base magnets ensure the connection bonds between

modules, while the edge magnets play a big rule in
module’s manipulation cause it’s free to rotate and due
Figure 3: components and sensors of block module
to the chamfered edges of modules, each module can
easily rotate about the other modules to its new position by using the strong bond provided by the
base magnets. The modules decide how, when and where to move so they'll take a large group of
cubes and form the required shape and deliver those instructions at a very high level and after that
the cubes decide on their own how to go about accomplishing that task as illustrated in (Fig. 1).[8]

Application:
The M-Blocks 2.0 is all new technology the has been revealed at October of 2019 from the
MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) , and this new release has
amazing ability of communicating cubes with each other and collaborate and do tasks on their
own and this gives them on future developments to be scaled to numbers of millions to do amazing
structure and tasks .

Page | 8
 Chain/tree architectures
Definition:
Chain architecture is a robot consists of modules connected with a string or tree topology.
These modules are versatile but difficult to control. One of the chain modular robots’ benefits is
providing many degrees of freedom that controls the whole-body locomotion.

General mechanism:
In the chain architecture, the reconfiguration scheme is achieved by detaching the
connection of an array of modules from one point of the architecture and re-attaching it at a
different point maintaining the direct connection of the entire assembly. This is achieved either
autonomously or manually under human control.

Example robot:
PolyBot:

PolyBot modules are based on an interface carrying the mating pins and holes that used to
connect modules together.

Components and mechanism:

The connection mechanism is that each cubical module has
two opposed interfaces, each face is designed with a set of four pins
and four holes arranged on concentric circles, 90° apart. During
docking, the pins from one module gets into the holes of the adjacent
module. A spring-loaded latching mechanism -actuated by shape
memory alloy (SMA) strings- rotates a scissor-like mechanism that
latches onto the pins locking them in place as illustrated in (Fig4).
This process can easily be reversed when detaching the two
modules. Detaching process is fulfilled by actuating the latching
mechanism in the opposite direction. Controlling the process of
Figure 4: example of PolyBot configuration
docking in a chain architecture is relatively hard, especially when
the chain consists of large number of modules due to accumulations
in positional error. In PolyBot, the problem was addressed by splitting the docking process into
three phases: long range, medium range and short range. The long-range phase aims to bring two
modules from any position in space near to the region of one another. Once the position sensors
embedded in the mating faces sense the closeness of one module to the another, the medium-range
phase starts to take an action. This phase aims to close the gap that separates the two modules with

Page | 9
the use of infrared sensing which determines the position and orientation of one module with
respect to the other in six-dimensional spatial coordinates. The final phase is only initiated when
the two modules are approximately close. Using only joint angle sensors and an open-loop
controller based on a kinematic representation of the modules, compliant alignment between the
pins of one module and the holes of the mating module is achieved. The pins are then guided to
mate with the holes, and the two are permanently locked together via the SMA-actuated latching
mechanism.[2]
.

Hybrid architectures
Definition:
Hybrid architecture is a system that can create sort of complex shapes using different
configurations of the same modules. Combining capabilities from both the chain and lattice
architectures has been proposed for modular robots providing whole body locomotion in a hybrid
architecture.

Example robots:
Molecubes

Components and mechanism:

Cubical modules were also used in modular robotics, in
particular for Molecubes. Each module is (10×10×10 cm) in
dimensions and split into two parts along an inclined plane
perpendicular to the cube’s long diagonal. The revolute joint that
connects both parts is actuated by an embedded motor that enables
one half-cube to swivel around the other. Each face of the cube
carries an electro-magnetic interface that enables it to dock with
adjacent modules to form the lattice assembly. as illustrated in
(Fig. 5).The polarity of these electromagnets can be changed
electrically to attract, repel or act as a neutral element to the
neighbouring module. The reconfiguration of a Molecubes’
assembly is achieved manually via a simulator and replicated on a Figure 5: modules of Molecubes
physical formation. Several mobility patterns can be demonstrated
with this manual replicator, such as a quadruped formation with multiple grippers.[5]

Page | 10
Application:
A company named FESTO has injected this type of modular
robot in industrial field by making robotic arm that can be assembled
and rearranged for variety of combinations to make a one robotic
arm can do a multiple tasks and the ability of Molecubes to rotate
360 degree on a inclined plane and the robotic arm consist of
multiple modules that give the robotic arm a very large working
space reach, the only drawback is the speed of motion of the
Molecubes while rotating. as illustrated in (Fig. 6). Figure 6: FESTO robotic arm

SuperBot

Components and mechanism:

SuperBot is built up by separate modules, each module
consists of two half-cubes which pivot around a central link. This
link allows the modules to rotate around three different axes as the
link itself is capable of rotating around its own axis. The rotations of
the modules around the axes are realized, each separately, by a DC
motor which is embedded inside the cubes alongside complementary
electronics and control boards.
To enable data sharing between the modules, docking
interfaces with electrodes are embodied on all three faces of each
half-cube. These interfaces are genderless, which means that any
docking face of one module can dock with any other face of an Figure 7: example of SuperBot configuration
adjacent module in all 90° orientation intervals. Using table-based
and hormone inspired distributed controllers, the control methodology in the case of the SuperBot
was executed, with a software hierarchy separating low level device-specific code from high-level
assignment driven routines. A special-purpose simulator was also developed to validate the
kinematics and dynamics of any desired formation on a virtual terrain prior to real-time
reconfiguration. With this control and simulation approach, SuperBot modules are among the few
WBL modular robots that were tested in a real environment, such as on sand dunes in uphill and
downhill manoeuvres replicating space terrain conditions.[3]
Application:
SuperBot is designed for NASA space exploration missions that has been early developed
in 2007 and still being developed and tested in real challenging environment. (Fig.7)

Page | 11
 Mobile configuration change (MCC)
Definition:
The category of mobile configuration change (MCC) refers to modular robots where
individual modules manoeuvre and interact with the environment independently, gather, and
physically connect to one another to change the group configuration and augment the capabilities
of a single module. This generally takes the form of a head-to-tail docking process.

General mechanism:
In this category, individual modules are self-contained and possess the sensing, computing
and actuating capabilities to move and operate individually. The locomotion patterns of MCC
modules are traditional and are implemented using typical mechanisms that enable efficient
mobility of individual modules such as wheels and treads.

Example robots:

MilliBots

Components and mechanism:

In this unique mechanism, the modules depend on shape-memory
alloy (SMA), which can be shaped but returns to its original form when
heated, and steel pins to reconfigure themselves. In the front of each module
exist two steel pins (male) which couple with a receptacle (female) placed
on the back of the other module, in a formation that resembles the train. as
illustrated in (Fig. 8) After that comes the role of the SMA wire, which
works as latching mechanism when triggered by a heating current, Figure 8: module of MilliBots

providing rigidity and more strength to the coupling. The module of a

Millibots system fits into a cube of side length 50 mm, consisting of a
parallelepipedal mobile platform with two belts. A different actuator is used
to run each belt solely, in order to allow the module, which carries sensors,
power supply and communication devices, different types of motion;
forward, backward, or turning. The coupling mechanism is further actuated
by a separate motor that provides revolute pitch motion of the pins and the
docking mechanism, enabling undulatory locomotion for a train formation, Figure 9: configuration of MilliBots
as show in the opposite figure (9).[4] climbing downstairs

Page | 12
SMORES-EP

Components and mechanism:

“SMORES” stands for Self-assembling Modular Robot for
extreme Shapeshifting, and “EP” refers to the Electro-Permanent
magnets the module uses for its connector. Each module has four
active rotational degrees of freedom, which are: pan, tilt and
left/right wheels, commonly named as LEFT DoF, RIGHT DoF,
PAN DoF and TILT DoF for convenience. The module is also
equipped with four connectors which are Left, Right, Pan and Tilt;
these connectors use an array of electro-permanent (EP) magnets.
LEFT DoF, RIGHT DoF and PAN DoF are free to rotate
continuously without angular limitations, to produce the twist Figure 10: the motion of SMORES-EP module
motion of docking ports relative to the rest of the module, while
TILT DoF is limited to ±90◦, producing a bending joint. LEFT DoF
and RIGHT DoF are also used as wheels in some applications, giving
differential drive locomotion to each single module. The module in
SMORES-EP is considered as a cube having four docking ports
named LEFT Face, RIGHT Face, TOP Face and BOTTOM Face for
convenience. Each one of these ports (faces) can form a strong bond Figure 11: several configurations of SMORES-EP
with other modules, or with metallic object, by using the electro-
permanent magnets that are arranged in a ring, with south poles counter clockwise of north. This
specific arrangement of the magnets is what makes the connector hermaphroditic, allowing it to
connect in four different configurations. Moreover, these connected EP-Faces can also exchange
data through the magnetic coupling of connected EP-magnets which are capable of UART
(Universal Asynchronous Receiver/Transmitter, a circuit in a microcontroller) serial
communication. (Fig. 10) (Fig. 11).[7]

Page | 13
 Future directions
As stated before, the modular robot faces a few major obstacles. These obstacles can be
summarized into four main challenges that the researchers and engineers are trying to overcome,
referring mainly to the hardware of the module and the algorithmic infrastructure of its technology.

1-Docking and coupling

The docking mechanism in a certain modular robot is what distinguishes it from other
robots, as it can solely define what kind of application the robot can be used for. The rigidity of
the connection realized by the docking mechanisms determines the maximum number of modules
that can be docked together in a certain formation without leading to unwanted disconnections or
any malfunction, especially due to the robot’s weight. If the connection achieved by the coupling
mechanism is not rigid enough, the robot might end up in being used only in laboratories for
research as it can perform a real-life task. While the ease and complexity of the docking mechanism
determine the robot’s application, having a complicated docking mechanism in a robot used to
rescue people will force the robot to take much time to reconfigure itself, which will probably
cause undesirable results.

2-Autonomy
Robots have always been about autonomy, requiring less information from the human to
execute the desired task, which saves both money and energy. For modular robots specifically,
where a huge number of temporary joints is found in the single structure, some aspects of
autonomy must be attained. Otherwise, a detailed and complex interface of a human operator
would be required to generate motion and perform tasks. However, it’s extremely complex,
difficult and expensive to make each module completely autonomous so several of them can build
one fully autonomous robot.

3-Field testing
This is one of the hardest obstacles facing the modular robots, as the existing technology
and materials used in the modular robot manufacturing do not fit the real environment and hard
terrains. It has been predominately reported that field testing is very limited to safe and well
controlled environments, such as laboratories, which is insufficient to demonstrate the
practicability and feasibility of the robot [6].

Page | 14
4-Modular manipulation
Most robots nowadays are provided with a manipulation system, which is one of the most
common ways for the robot to interact with the surroundings. However, the main task of the
modular robot is using specific movement to change its structure or shape, letting manipulation
become a secondary task, which is why not all modular robots can combine between both
locomotion and manipulation in the same morphology. Not providing the modular robot with a
gripper or a manipulator arm limits the robot interaction with the surrounding to only receive
information and act based on the given data, but to act upon the surroundings.

Availability and price

Because of the fact that modular robots are relatively
new technology and depend on development of other
technologies specially electronics most of the modular robots
that mentioned earlier are only on test and development in
laboratories and they are not available in market except a
child game called Cubelets (Fig. 12) and available in market
for price of set 250$.

Figure 10: cubelets group set

Page | 15
 Conclusion
The invention of modular robots, a machine that can change its form to do different tasks while
being able to move in different configurations as well as reacting to the surroundings using sensors,
cameras and manipulating systems, is astonishing. This type of robots is quite promising, having
the potentials to be used in almost all fields, using various docking methods and locomotion
systems, limited only by the human imagination. However, much research is still needed to
overcome the obstacles facing the spreading of modular robots; docking methods, autonomy, field
testing and manipulation. Many institutes, universities and laboratories are already taking steps
that help the progress of modular robots and overcoming the barriers, yet, more effort is necessary.
And just like computers and industrial robots, it is expected that once the modular robots spread
out, it will ignite a new industrial revolution, and many of our daily tasks will be carried out by
modular robots, since it will be easy to fill our homes, cars and even our pockets with a few
modules that can execute miscellaneous tasks. Right now, all it needs is to be shown more interest
by different establishments, and to be given more effort in order to arise.

Page | 16
 References

[1] V. Zykov, E. Mytilinaois, M. Desnoyer, H. Lipson, Evolved and designed self-reproducing

modular robotics, IEEE Transactions on Robotics 23 (2) (2007) 308–319.
[2] M. Yim, New locomotion gaits, in: Proceedings of the IEEE International Conference on
Robotics and Automation, ICRA’94, 1994, pp. 2508–2514.
[3] W.-M. Shen, M. Krivokon, H. Chiu, J. Everist, M. Rubenstein, J. Venkatesh, Multimode
locomotion via SuperBot reconfigurable robots, Journal of Autonomous Robots 20 (2) (2006) 165–
177.
[4] H.B. Brown, J.M. Vande Weghe, C.A. Bererton, P.K. Khosla, Millibot train for enhanced
mobility, IEEE/ASME Transactions on Mechatronics 7 (4) (2002) 452–461.
[5] D.J. Christensen, J. Campbell, K. Stoy, Anatomy-based organization of morphology and
control in self-reconfigurable modular robots, Journal of Neural Computing and Applications 19
(6) (2010) 787–805.
[6] K. Stoy, H. Kurokawa, Current topics in classic self-reconfigurable robot research, in:
Proceedings of the IROS Workshop on Reconfigurable Modular Robotics: Challenges of
Mechatronic and Bio-Chemo-Hybrid Systems, 2011, San Francisco, CA, USA
[7] Satoshi Murata et al. M-tran: Self-reconfigurable modular robotic system. Mechatronics,
IEEE/ASME Trans., 2002.
[8] J. W. Romanishin, K. Gilpin, and D. Rus, “M-blocks: Momentumdriven, magnetic modular
robots,” in Intelligent Robots and Systems (IROS), 2013 IEEE/RSJ International Conference on.
IEEE, 2013, pp. 4288–4295

Page | 17