Metamorphic Robots PDF

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/2683201

Design And Implementation Of Metamorphic Robots

Article · October 1996


Source: CiteSeer

CITATIONS READS
144 5,981

3 authors, including:

David Stein Gregory S. Chirikjian


Johns Hopkins University Johns Hopkins University
7 PUBLICATIONS   404 CITATIONS    372 PUBLICATIONS   9,013 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Hopkins Library Robot View project

Protein kinematics View project

All content following this page was uploaded by Gregory S. Chirikjian on 22 December 2014.

The user has requested enhancement of the downloaded file.


Proceedings of
The 1996 ASME Design Engineering Technical Conference and
Computers in Engineering Conference
August 18-22, 1996, Irvine, California

96-DETC/MECH-1149

DESIGN AND IMPLEMENTATION OF METAMORPHIC ROBOTS

Amit Pamecha1 Chih-Jung Chiang2


Department of Mechanical Engineering Department of Mechanical Engineering
Johns Hopkins University Johns Hopkins University
3400 N. Charles Street 3400 N. Charles Street
Baltimore, Maryland 21218 Baltimore, Maryland 21218
[email protected] (410)516-4573
[email protected]

David Stein3 Gregory Chirikjian4


Department of Mechanical Engineering Department of Mechanical Engineering
Johns Hopkins University Johns Hopkins University
3400 N. Charles Street 3400 N. Charles Street
Baltimore, Maryland 21218 Baltimore, Maryland 21218
(410)516-7127
[email protected]

ABSTRACT 1 INTRODUCTION
This paper discusses issues in the design and implementa- A metamorphic robotic system Ch94] is a collection
tion of metamorphic robotic systems. A metamorphic robotic of independently controlled mechatronic modules, each of
system is a collection of independently controlled mechatronic which has the ability to connect, disconnect, and climb over
modules, each of which has the ability to connect, disconnect, adjacent modules. Each module allows power and informa-
and climb over adjacent modules. A metamorphic system can tion to ow through itself and to its neighbors. A change in
dynamically recongure by the locomotion of modules over their the metamorphic robot morphology (i.e., a change in the rel-
neighbors. Thus they can be viewed as a collection of connected
modular robots which act together to perform the given task. ative location of modules within the collection) results from
The planar metamorphic robots described in this paper consist the locomotion of each module over its neighbors. Thus
of hexagonal or square modules. Because of their shape, the a metamorphic system has the ability to dynamically self-
modules completely ll the plane without any gaps, their centers recon gure.
forming a regular lattice. Both the hexagonal and square mod- Metamorphic systems can be viewed as a large swarm
ules are provided with electromechanical coupling mechanisms (or colony) of connected robots which collectively act as
actuated by D.C. motors. These connectors help to couple and a single entity. What distinguishes metamorphic systems
uncouple modules as they move around each other to form dif- from other recon gurable robots is that they possess all of
ferent congurations. The modules are currently controlled by the following properties:
an external processor.
1. All modules have the same physical structure, and each
must have complete computational and communication
1 Graduate Student functionality. This allows uniform treatment of mod-
2 Graduate Student ules in the planning problem.
3 Undergraduate Student
Assistant Professor
Copyright 
4
1 c 1996 by ASME
2. Symmetries in the mechanical structure of the modules design of planar square modules and their locomotion pro-
must be such that they ll planar and spatial regions cess is discussed. One important aspect of such design is
with minimal gaps. In this way, a lattice of modules is that the centers of the square modules form a Cartesian
formed for any task. lattice, thus can be easily extended to spatial case with a
3. The modules must have enough degrees of freedom to cubic module design. Section 6 presents the conclusions.
be able to `walk' over adjacent modules so that they
can recon gure without outside help.
4. Modules must adhere to adjacent modules, e.g., there
must be electromechanical or electromagnetic connec- 2 LITERATURE REVIEW
tors between modules which can carry load. This causes The idea of a metamorphic robotic system diers from
the collection of modules to act as a single physical ob- related concepts presented in the literature. Three types of
ject. modular recon gurable robotic systems have been proposed
in the literature: (1) robots in which modules are recon g-
Potential applications of metamorphic systems com- ured using external intervention, e.g., BeZL89, CoLDB92,
posed of a large number of modules include : (1) obsta- Sci85, Wu86] (2) cellular robotic systems in which a het-
cle avoidance in highly constrained and unstructured envi- erogeneous collection of independent specialized modules
ronments (2) `growing' structures composed of modules to are coordinated, e.g., Be88, BeW91, FuN88, FuK90] (3)
form bridges, buttresses, and other civil structures in times swarm intelligence in which there are generally no physical
of emergency (3) envelopment of objects, such as recov- connections between modules, e.g., HaB91]. Most recently,
ering satellites from space (4) Performing inspections in two other types of modular recon gurable robotic systems
constrained environments such as nuclear reactors. Some of have been considered. Yim93,Yim94] considered modular
these applications are shown in Figure 1. robots composed of a few basic elements which can be com-
posed into complex systems, and used for various modes
of locomotion. MuKK94, MuKK95] considered a `fractal'
system composed of modules with zero kinematic mobility,
but which can `walk' over each other in discrete quanta due
to changes in the polarity of magnetic elds.
In the present work, where the design of a mechanically
error tolerant coupling mechanism is also important, an-
other body of literature is relevant. Namely, work that deals
Figure 1. Examples of Metamorphic Robot Applications with the mechanics of pushing and friction and work that
uses geometric and physical constraints to guarantee desired
performance with minimal numbers of crude sensors, e.g.,
AkM92, CaG94, ErM88, Ma93, PeS89, PeBG93]. By using
This paper discusses issues in the design of two dier- this `minimalist' philosophy, we have developed a coupling
ent planar modules. Section 2 contains a brief literature device for metamorphic robots that requires no sensors and
review. In Section 3 the basic design of planar hexagonal has a geometry that allows for signi cant errors. This is
modules and their locomotion process is discussed. One of important because a metamorphic robotic system may fre-
the most important aspects of the above design is the con- quently need to recon gure, and so the connection between
nector mechanism. The connector not only has to adhere modules must be reliable.
to the adjacent module but also has to allow sliding motion The motion planning problem of a metamorphic robot
of the links. In addition it should be a passive or a quasi- has been previouly discussed by Ch94, PECh95]. SMCh93]
passive mechanism so as to use minimum power i.e, once it consider the motion of planar hexagonal metamorphic mod-
has coupled or uncoupled the modules, it should require no ules using 'local' constraints and describe the performance
power to maintain the connection. This paper describes one of various algorithms under these constraints.
such error-tolerant coupling mechanism design for hexago-
nal modules, i.e the coupling mechanism can take into ac-
count wide variations in connector position on the mating
link. Section 4 describes some alternatives in the design of 3 BASIC DESIGN AND LOCOMOTION OF HEXAGONAL
couplings for hexagonal modules and discusses the design MODULES
and implementation of an error-tolerant connector mecha- This section describes the design of planar hexagonal
nism satisfying the above conditions. In Section 5 the basic modules and the locomotion process of the modules.
2 Copyright 
c 1996 by ASME
3.1 Module Design
Section 1 described the four important properties of
metamorphic robots. One of the designs which satis es all
those properties in the planar case involves the use of hexag-
onal modules. Each module, as shown in Figure 2, consists
of six links of equal length forming a six bar linkage.

Figure 3. Polarity matching in the reconguration of metamorphic


robot.

motion and issues signals to the actuators. It is also as-


Figure 2. Design of a planar hexagonal module. sumed that eventually power and information between mod-
ules will be transmitted through the coupling mechanism.
Because of the hexagonal shape, the modules com-
pletely ll the plane without any gaps. The centers of 3.2 The Locomotion Process
the hexagonal modules form a regular lattice and thus each There are three ways to view the macroscopic motion
module can be treated as part of a lattice structure. Each of the modules of a metamorphic robot: (a) Single module
hexagonal module can be viewed as a closed six bar link- motion, which involves the motion of one, and only one,
age with three degrees of freedom which are controlled by module per time step from one lattice point to another (b)
placing actuators at alternate joints of the module. This Motion involving two or more modules, moving together or
enables each module to move around another while remain- separately at each time step (c) Fixed morphology motion
ing connected at all times during this motion. The modules in which the connection between the modules remains the
are provided with electromechanical connectors or coupling same, and a change in con guration occurs by changing the
mechanism actuated by D.C. motors. The connectors serve joint angles of the modules. The present work discusses
the important purposes of coupling adjacent modules, fa- issues involving the rst type of motion (single module mo-
cilitating power and information ow and allowing sliding tion).
motion of links to enable the module to recon gure. In ad- The recon guration of metamorphic robots with single
dition, they have to be passive or quasi-passive mechanisms module motion takes place by the locomotion of modules
and error tolerant to take into account variation in module around each other while remaining connected to each other
position. Each module carries male and female connectors at all times. This can also be described as the `rolling' of
or dierent polarities on alternate links. Because of the one module over others. Figure 4 illustrates the locomo-
symmetry of the module, male connectors always meet fe- tion procedure. Observe that the module which is moving
male connectors and vice-versa as illustrated in Figure 3. remains connected to the other module during the entire
In addition this symmetry is maintained over the entire process. For clarity, let's de ne the module which is mov-
structure, i.e. the adjacent links on the boundary of the ing at a given time to be mobile and the module over which
collection of all modules are of dierent types or of dierent it moves to be xed. The motion of the mobile module is
polarities. The connector mechanism is one of the most im- achieved by controlling the three degrees of freedom of the
portant aspect of module design and is explained in greater module by three actuators on alternate joints. For a de-
detail in Section 4. tailed description of the kinematics of the mechanism and
Each module must also contain a microprocessor which how joint angles are altered see Ch94].
controls the link actuators and the connector motors mak- Any implementation of the locomotion process de-
ing the module computationally self-contained. However, scribed above in practice has problems because of the -
presently the modules are controlled by an external mo- nite thickness of the links. Due to the nite thickness, the
torola 68HC11 controller which determines the direction of axes of rotation of the joints of two mating links are not
3 Copyright 
c 1996 by ASME
xed and the mobile modules move together so that the old
connection and the new one are parallel to the mating links,
i.e. each of the two mating links moves by 60o towards each
other instead of the link of the mobile module moving by
120o. This ensures proper alignment but requires a coor-
dinated simultaneous movement of other modules in the
structure.
The above argument indicates that the locomotion pro-
cedure of modules warrants a mechanism that somehow re-
moves the displacement between the links. Also, since the
connectors do not meet at the same position relative to the
links, an error tolerant connector mechanism is required.
The next section discusses some of the alternatives which
overcome the above problems.
Figure 4. The locomotion of one module around another. One of
the links of the mobile module always remains connected to the xed
module. 4 DESIGN OF CONNECTOR MECHANISM
As described in the previous section, the locomotion of
modules necessitates the use of an error tolerant connector
design which also aligns the mating links of the modules.
A number of possible designs can satisfy the above require-
ments. This section rst describes some of the possible al-
ternatives and then describes the design actually used for
overcoming the limitations of the earlier designs.

4.1 Alternatives in Connector Design


One alternative for module locomotion is to make the
links extensible, i.e. the links can contract or expand rel-
ative to their normal length. In this case the mating link
on the mobile module extends (Figure 6, step 1) aligning
itself to the link of the xed module (Figure 6, step 2),
locks in, releases the old connection and contracts to regain
Figure 5. Displacement between the links due to nite link thickness the normal shape (Figure 6, step 3). The obvious problem
with this method is that it needs an actuator for each link
to extend or contract in addition to the actuators required
for the motion of the joints and for locking unlocking of
coincident. As a result, when a module moves around the connectors.
other, the links in the new position are not aligned. For Another approach which tries to solve the two problems
an illustration of this see Figure 5. The displacement (d) is one in which the connector is a spring loaded mechanism
between the two links is a function of the width (w) of the which aligns the two links together. For an illustration of
links and is given as d = w tan  where  is 30o. This can be this see Figure 7. The connector on the link of the mobile
derived very simply by observing the module geometry in module slides into a wedge shaped connector on the oppos-
Figure 5. If this displacement or misalignment is not elimi- ing link by compressing the spring. The relative compres-
nated, the next motion of the module in the same direction sion and extension of the springs on the links is given by
cannot take place. Another problem due to this misalign- the minimization of the strain energy function associated
ment and due to the possibility of motion in both clockwise with the springs. When the old connection gets released
and counterclockwise directions is that the connectors on the spring aligns the entire module to normal position. The
the two opposite links do not meet at an exact point on problem with this method is that for moving the entire mod-
the links, i.e. the connectors themselves get displaced with ule once the old connection gets released, a sti spring is
respect to each other. required which in turn requires a high torque actuator to
An alternative strategy for motion is one in which the act on it in the rst place. As a result this design is dicult
4 Copyright 
c 1996 by ASME
Figure 8. Module motion using electromagnetic coupling.

to magnetic force. The rigid connection between the mod-


ules is provided by the magnetic force between the two con-
nectors . A similar scheme has been used by MuKK94,
MuKK95] in the development of self-assembling machines
or `fracta'. The problem with this approach is that connec-
tion between dierent modules are active instead of passive.
Figure 6. Motion involving link extension and contraction. This means that a large amount of power for the electromag-
nets is required at all times to keep the modules connected.

4.2 Error Tolerant Coupling Mechanism


The design used for the current modules consist of an
error tolerant connector which tries to overcome the limita-
tions of the above designs while providing proper alignment
and a passive connection between the modules. The mech-
anism takes into account the nite link thickness and helps
in carrying out the motion as described in Section 3.
The coupling mechanism consists of two dierent types
of connectors, referred to as male and female. Let's de ne
the link carrying the male connector as link A and the one
carrying the female connector as link B. See Figures 9 and
10.
Link A consists of two parallel plates with space in be-
tween. The space carries a T-shaped protrusion mounted
on a sliding mechanism. The protrusion is held in place by
two springs, one on each side (see Figure 9). As a result,
Figure 7. The spring loaded connector shown for two links from each when force is applied to the protrusion, it can slide sideways
module. by compressing the springs.
Link B also consists of two parallel plates with space
in between. The space carries two cams, able to rotate
to implement and increases the power requirement. 120o about their axes. The cam dimensions are such that
Yet another option is to use electromagenetic connec- when they move in, they completely lock the protrusion on
tors which serve the dual purpose of aligning the two mating the corresponding link of the other module. The locking of
links and connecting the two modules. One design employ- the protrusion by the cams prevents any lateral movement
ing such a method is shown in Figure 8. Each link has an while the T-shaped structure of the protrusion stops any
electromagnet as a connector. When the old connection is longitudinal movement. The cams are operated in unison
released (by repulsing the electromagnets forming the con- by a single actuator (a small DC motor in our case) which is
nection), the electromagnets align the links completely due connected to the cams by a set of 4 gears as shown in Figure
5 Copyright 
c 1996 by ASME
The compressed spring on A1 now aligns link A1 and B2
completely. This completes one move of the module.

Figure 9. Semi-exploded view of link A showing the spring loaded


protrusion.

Figure 11. Motion sequence demonstrating action of the error tolerant


coupling mechanism.

As can be seen, only one actuator is needed for each


Figure 10. Semi-exploded view of link B showing the double cam coupling mechanism i.e. one actuator for every two links
mechanism. Shown here without the actuator
of the module. The links are aligned exactly by the use of
springs on alternate links. A large space is available between
10. As a result the cams open and close simultaneously. the cams(in their open position), as a result the protrusion
The motion sequence utilizing this mechanism is shown or the connector on the opposing link can mate from both
in Figure 11, steps 1 to 4. Step 1 shows the original po- directions and need not mate at an exact position, i.e. the
sition of the two adjacent modules, one of which is about coupling mechanism is error tolerant. Another advantage
to move around the other. In step 2 link A1 of the mobile of the above design is that the locking mechanism is passive
module moves towards link B2 without displacing link B1. in the stationary state and so uses very little power.
In step 3 this motion is continued, but as link A1 moves A working prototype using the above design is shown
into B2, it slides the connector on link A2 by compressing in Figure 4. For a video of the locomotion process involving
one of the springs. As a result, link B1 gets displaced from error tolerant connectors, see PCh95].
its original position and shifts sideways. The cams on B2
now close in, aligning link A1 parallel to B2. This results
in the connector on A1 getting displaced from it's mean po-
sition and one of the springs getting compressed. In step 5 BASIC DESIGN AND LOCOMOTION OF SQUARE MOD-
4 the cams on link B1 open up, releasing the connector on ULES
A2. B1 then rotates by 120o and the structure attains the This section describes the design of square modules and
con guration shown in the gure corresponding to step 4. the locomotion process of the modules.
6 Copyright 
c 1996 by ASME
5.1 Module Design
Another design which satis es all the properties men-
tioned in Section 1 involves the use of square modules. The
square modules can completely ll the plane without any
gaps as do the hexagonal modules. The centers of the square
modules form a Cartesian lattice and each module can be
viewed as part of a lattice structure. The fact that the
centers of the square modules form a Cartesian lattice is
important since we can easily adapt this concept and ex-
pand it into spatial case with a cubic module design. Un-
like a hexagonal module, which has the required kinematic
degrees of freedom to `roll' over neighboring modules by
changing its joint angles, a square module needs connect-
ing mechanisms to couple adjacent modules and enable the Figure 13. Polarity matching in the robot
module to `walk' over neighboring modules by sliding.
Each module, as shown in Figure 12, carries male or
female connectors, or dierent polarities, on each of its
edges. Because of the symmetry of the modules, the lo-
comotion always results in edges with opposite polarity or
male/female connectors meeting with each other. This sym-
metry is maintained over the entire structure as illustrated
in Figure 13.

Figure 14. Male connector and the shuttle

connectors.
The female connector consists of three parallel plates
with seven gears mounted on the middle plate. The two
Figure 12. The mechanical structure of a square module outermost gears mesh with the rack of the shuttle of its
neighboring male connector. One of these two gears will be
actuated by a D.C. motor and serves as a driving gear for
the female connector. The top and bottom plates, called
Figures 14 and 15 are semi-exploded views of the male the jaws, can slide open or close along three guiding rods
and female connectors respectively. perpendicular to the plates. This opening and closing jaw
The male connector consists of two parallel plates with movement is actuated by a D.C. motor. The jaws are to
space in between. There are one driving gear and two rollers remain closed (see Figure 16) unless there is another mod-
mounted on the bottom plate. An H-shaped mating link ule coming from the direction perpendicular to the edge of
called the shuttle with racks attached to both sides are to the female connector. In that case, we have to open the
be driven by the driving gear. The shuttle will remain con- jaws (see Figure 17) so that the shuttle of the incoming
nected to the male connector while sliding back and forth module can come in. The shuttle will be locked in position
in the rails formed by the tracks of both male and female when the jaws close, thus completing the mating process.
7 Copyright 
c 1996 by ASME
Figure 17. Open position for the jaws

each other while remaining connected to each other at all


times. There are two types of such `sliding' motion: (a)
Vertical or horizontal motion, which involves the motion of
a mobile module in the vertical or horizontal direction (b)
Diagonal motion, which involves the motion of a moving
module in the diagonal direction.
The locomotion procedure for vertical or horizontal mo-
Figure 15. Female connector tion is rst to identify the mating pair of connectors and ac-
tuate the driving gear of the female connector to move the
module in the desired direction. Once the mobile module
is making the initial connection with the new neighboring
module, the driving gear of the female connector in the new
mating pair of connectors is actuated until the mobile mod-
ule reached the designated position.
The motion sequence demonstrating the locomotion
procedure for diagonal motion is shown in Figure 18, steps 1
to 5. Step 1 shows the original position of the two adjacent
modules, the left one, the mobile module, is about to move
to its diagonal direction to the top of the xed module. In
step 2 the driving gear of the female connector drives the
Figure 16. Closed position for the jaws mobile module halfway up then the driving gear of the male
connector takes over and moves the mobile module one full
module distance up. In step 3 the driving gear of the male
In the case of releasing the connection, we have to follow connector in the mobile module slides the shuttle one half
the above sequence in reverse order. A total of six D.C. of a module distance to the right to make connection with
motors are needed in a single module, two each for oper- the female connector for the xed module. In step 4 the
ating the driving gears in the male connectors, the driving connection between the rst pair of connectors releases and
gears in the female connectors, and the jaws opening/closing the shuttle moves back up to its normal position. In step 5
mechanisms. There should be some oset distance between the driving gear for the male connector brings the shuttle
the connectors along the vertical and horizontal directions back to its normal position while the driving gear for the
to avoid the shuttle interference. This will be discussed in female connector moves the mobile module further right to
more detail in the next subsection. the nal position. This completes one diagonal move of the
module.
5.2 The Locomotion Process A prototype using the above design is shown in Fig-
The recon guration of metamorphic robots with square ure 19. A stamp-sized Basic microprocessor is used for each
modules takes place by the locomotion of modules around module to control the motion sequence.
8 Copyright 
c 1996 by ASME
Figure 19. Hardware illustration of two mating square modules

the design of such modules is the coupling mechanism. In


the hexagonal module design, an error tolerant coupling
mechanism was discussed and the resulting locomotion pro-
cess demonstrated. In the square module design, a cou-
pling mechanism consisting of a shuttle and male/female
connectors was discussed and the resulting locomotion pro-
cess demonstrated.

7 ACKNOWLEDGEMENTS
This work was made possible by a 1994 Presidential
Faculty Fellow Award from the NSF.

8 REFERENCES
AkM92] Akella, S., Mason, M.T., \Posing polygonal ob-
jects in the plane by pushing," Proc. IEEE 1992 Inter-
national Conference on Robotics and Automation, pages:
2255-2262.
BeZL89] Benhabib, B., Zak, G., Lipton, M.G., \A Gener-
alized Kinematic Modeling Method for Modular Robots,"
Journal of Robotic Systems, vol. 6, No. 5, 1989, pp. 545-
571.
Be88] Beni, G., \Concept of Cellular Robotic Systems,"
IEEE Int. Symposium on Intelligent Control, Arlington,
VA, August 24-26, 1988
BeW91] Beni, G., Wang, J., \Theoretical Problems for the
Realization of Distributed Robotic Systems," 1991 IEEE
Figure 18. Motion sequence demonstrating the diagonal transforma- Conf. on Robotics and Automation, 1991, pp. 1914-1920.
tion CaG94] Canny, J., Goldberg, K., \A RISC approach
to robotics," IEEE Robotics and Automation Magazine,
pages: 26-28, March 1994.
6 CONCLUSION Ch94] Chirikjian, G.S., \Kinematics of a Metamorphic
Robotic System," Proceedings of the 1994 IEEE Int. Conf.
In this paper we described the design and implementa- on Robotics and Automation, San Diego, CA, May 1994,
tion of metamorphic robots consisting of planar hexagonal pp.449-455.
or square modules. One of the most important issues in CoLDB92] Cohen, R., Lipton, M.G., Dai, M.Q., Benhabib,
9 Copyright 
c 1996 by ASME
B., \Conceptual Design of a Modular Robot," Journal of
Mechanical Design, March 1992, pp. 117-125.
Di59] Dijkstra, E. W., \A note on two problems in Con-
nexion with Graphs," Numerical Mathematics, 1, 1959, pp
269-271.
ErM88] Erdmann, M.A., Mason, M.T., \An exploration
of sensorless manipulation," IEEE Journal of Robotics and
Automation, Vol.4, No.4, August 1988.
FuN88] Fukuda, T., Nakagawa, S., \Dynamically Recon-
gurable Robotic Systems," Proc. 1988 IEEE Conf. on
Robotics and Automation, pp. 1581-1586
FuK90] Fukuda, T., Kawauchi, Y., \Cellular Robotic Sys-
tem (CEBOT) as One of the Realization of Self-organizing
Intelligent Universal Manipulator," Proc. 1990 IEEE Conf.
on Robotics and Automation, pp. 662-667.
HaB91] Hackwood, S., Beni, G., \Self-organizing Sensors
by Deterministic Annealing," IROS'91, Osaka, Japan, pp.
1177-1183.
Ma93] Mason, M.T., \Kicking the sensing habit," AI Mag-
azine, pages 58-59, Spring 193.
MuKK95] Murata, S., Kurokawa, H., Kokaji, S., \Self-
Organizing Machine," Video Proceedings, 1995 IEEE Int.
Conf. on Robotics and Automation, Nagoya, Japan, May
1995.
PCh95] Pamecha, A., Chirikjian, G.S., \A Metamorphic
Robotic System: Simulation and Connector Mechanism
Demonstration," Video Proceedings, 1995 IEEE Int. Conf.
on Robotics and Automation, Nagoya, Japan, May 1995.
PECh95] Pamecha, A., Ebert-Upho, I., Chirikjian, G.S.,
\Some Useful Metrics for Modular Robot Motion Plan-
ning," JHU Tech. Report RMS-95-10, Johns Hopkins Uni-
versity, 1995
PeS89] Peshkin, M.A., Sanderson, A.C., \Planning robot
manipulation strategies for workpieces that slide," IEEE
Journal of Robotics and Automation, Vol.4, No.5, October
1988.
PeBG93] Peshkin, M.A., Brokowski, M., Goldberg, K.,
\Curved fences for part alignment," Proc. IEEE 1993 In-
ternational Conference on Robotics and Automation, pages:
467-473.
Sci85] Sciaky, M., \Modular Robots Implementation,"
Handbook of Industrial Robotics, S. Nof, ed., John Wiley
and Sons, New York, 1985, pp. 759-774.
SMC93] Seabury, S.D., Malik, R., Chirikjian, G.S., \Com-
parisons of Local Heuristics for Motion Planning of a Self-
Recon gurable Robot", JHU Tech. Report RMS-93-3,
Johns Hopkins University, 1993.
Wu86] Wurst, K.H., \The Conception and Construction of
a Modular Robot System," Proceedings of the 16th Interna-
tional Symposium on Industrial Robotics, pp. 37-44, ISIR,
1986
Yim93] Yim, M., \A Recon gurable Modular Robot with
Many Modes of Locomotion," Proceedings of the 1993
JSME International Conference on Advanced Mechatronics,
Tokyo, JAPAN, August 1993, pp. 283 - 288.
Yim94] Yim, M., \New Locomotion Gaits," Proceedings of
the 1994 IEEE Int. Conf. on Robotics and Automation,
San Diego, CA, May 1994.

10 Copyright 
c 1996 by ASME

View publication stats

You might also like