Motion and Modular Architecture

ABSTRACT
Simon Kim This paper presents an implementation of an architectural module that corresponds
University of Pennsylvania to a long serial chain modular robot. As such, this configuration poses possibilities
that can move using travelling wave gaits based on snakes and caterpillars. The
Mark Yim g a i t s a r e c o n t r o l l e d w i t h a G a i t C o n t r o l Ta b l e w h i c h i s a s i m p l e b u t p o w e r f u l
University of Pennsylvania way to coordinate the motion of a multiple degree-of-freedom systems. The gaits
are implemented on a self-sufficient modular reconfigurable robot with onboard
Jedtsada Laucharoen p o w e r, c o m p u t a t i o n , s e n s o r s a n d a c t u a t o r s .
University of Pennsylvania

Michael Wetmore
University of Pennsylvania

Sanam Salek
University of Pennsylvania

Sam Pan
University of Pennsylvania

350

acadia 2011 _proceedings integration through computation

1 I n trodu c tion

1.1 MODULAR ARCHITECTURE

T h i s d e f i n it i o n may b e s ep arat ed t o t wo l evels of orde r, modu larity in arc h ite c tu re

r e f e r r i n g t o i nt eg rat i o n o f b ui l d i ng co mp o nent s for e ase of man u fac tu re , or to large r
m o d u l e s o f o ccup ancy t hat are mo re freq uen tly re fe rre d to as Mat Bu ildin gs (Le
C o r b u si e r 19 6 4 ; Smi t hs o n 1 9 7 4 ; Al l en 2 0 0 1 ; Sarkis 2001). Wh ile th e se Mat Bu ildin g
p r o j e c t s p ro mo t e an i d ea o f fl exi b i l i t y and i nd e te rmin ac y, little h as be e n atte mpte d in
t h e d e v e l o pment o f d ynami c res p o ns e, es p eci a lly in c on figu ration , an d in move me n t.

1.2 MODULAR ROBOTICS

L o n g se r i a l chai ns o f mo d ul ar ro b o t s wi t h man y de gre e s of fre e dom h ave a varie ty

o f a p p l i c a ti o ns i ncl ud i ng s nake-l i ke l o co mo t i on wh ic h h as be e n e xte n sive ly stu die d
( F u k u s h i m a and Hi ro s e 1 9 9 0 ; Chi ri kj i an and B urdic k 1995; Y im, D u ff an d Ru fas, 2000).
H o w e v e r, ap p l yi ng t hes e p ri nci p l es t o an arch ite c tu re th at is mobile h as be e n le ss
evident.

T h i s p a p e r p res ent s t wo mo d es o f mo t i o n us in g Modu lar Arc h ite c tu re re -in te rpre te d

a s M o d u l a r R o b o t s t o i nves t i g at e a hi g her o rd e r of re c on figu ration be yon d th e h istoric
t y p e a c c o r d i ng t o d ynami c s cenari o s o f o ccup an c y, as we ll as loc omotion . T h e forme r
r e q u i r e s l o cal mo vement co ns t rai ned al o ng a n axis, an d th e latte r propose s global
m o v e m e n t o f s nake-l i ke archi t ect ure al o ng t errain . T h e se move me n ts are produ c e d by
G a i t C o n t r o l Tab l es t hat co o rd i nat e i nd i vi d ual modu le s.

2 Modu la r Arc h ite c tu re Pre c ide n ts

2.1 MAT BUILDINGS

A l i s o n S m it hs o n (2 0 0 1 ) d efi ned a t yp e o f co nstru c tion syste m of fle xible c olle c tion

a n d o c c u p at i o n: “Mat b ui l d i ng can b e s ai d t o e pitomize th e an on ymou s c olle c tive ;
w h e r e t h e funct i o ns co me t o enri ch t he fab ri c an d th e in dividu al gain s n e w fre e doms
o f a c t i o n t h ro ug h a new s huffl ed o rd er, b as ed on in te rc on n e c tion , c lose kn it patte rn s
o f a s s o c i a ti o n and p o s s i b i l i t i es fo r g ro wt h, d i min u tion an d c h an ge .”

H e r r a t i o n al e exp l ai ns t hat , “To d ay s p ace i s total an d soc ie ty is u n ive rsal. T h e se

r e a l i t i e s m u s t b e refl ect ed i n o ur p l anni ng and bu ildin g. T h e u n de rstan din g mu st c ome
t h r o u g h t h e p ercep t i o n o f t he p art s , as t he whole syste m c an n e ve r be se e n . We mu st
d i sp e n se wi t h t he us e o f s ymb o l s and mo numents, for th e c e n tu r y h as c ast aside th e se
c r u t c h e s of aut ho ri t y. Ind eed i f aut ho ri t y can be said to e xist it c an on ly be th rou gh
c o n s e n t a nd has no need o f fo rmal i s m o r o f alle gorie s to impose itse lf.” T h e au th or
t h e n p r o c e ed s t o fo rmal i ze i t s rul es : t o t al o rd e r c oordin ate d to a sin gle artic u lation of
f u n c t i o n , i n a s p eci fi ed ‘ s t em-t o -cl us t er ’ o rg an ization .

2.2 BUILDING MODULES

W i t h i n t h e o p en framewo rk o f mat urb ani s m , modu le s c an be arran ge d in u se fu l

p a t t e r n s o f s i ng l e p i eces as wel l as l arg er ag g re gate s. T h e se modu le s are disc re te in
e n c l o s u r e and s p ace, and meant t o b e co ns t ruc te d se parate ly to be de ploye d on site .
F u r t h e r m o r e, t he mo d ul es are i d ent i cal i n s i ze an d mate rial, re su ltin g in in e xpe n sive
m a n u f a c t u ri ng d ue t o eco no mi es o f s cal e and allowin g for e asy c on n e c tion with on e
a n o t h e r. What S mi t hs o n ap p l i es b eyo nd t he sin gle modu le to ove rall syste m patte rn
i s a d e t e r m i ni ng fact o r o f s p ace as a funct i o n of time : “T h e syste ms will h ave more
t h a n t h e u sual t hree d i mens i o ns . They wi l l i ncl ude a time dime n sion . T h e syste ms will
b e s u f f i c i e nt l y fl exi b l e t o p ermi t g ro wt h and ch an ge with in th e mse lve s th rou gh ou t th e
c o u r se o f t hei r l i ves . The s ys t em wi l l remai n o pe n in both dire c tion s, i.e ., in re spe c t to
s m a l l e r s y st ems wi t hi n t hem as wel l as i n res p e c t to gre ate r syste ms arou n d th e m. T h e
s y st e m s w i l l p res ent , i n t hei r b eg i nni ng , an even ove rall in te n sity of ac tivity in orde r n ot
t o c o m p r o mi s e t he fut ure.”

This proposal of mat-urbanism can be viewed as congruent with the premise of

Modular Robotics.

351

kinetics , matter and computation

 3 Mo dul ar Ro bo ti cs

3.1 PRECEDENT

A modular robot is a versatile system consisting of many simple modules that can change their
configuration to suit a given task. These systems are inherently robust due to their redundancy,
adaptability, and ability to self-repair. ModLab, University of Pennsylvania
Fig. 1

The research in self-reconfigurable modular robots yields small-scale groups of individual

pieces and intelligence. These modules, when joined together, can provide not one overall
form but varying forms to suit a particular scenario. Common projects demonstrate a
walking robot made of self-similar parts that can show linear snake-like movement, can join
into a closed loop that rolls, or can reconfigure into four-legged or two-legged assemblies.
This modular re-grouping allows for an overall organization that can dynamically change in
response to need. Other robots of fixed shape, such as a humanoid, are designed for one
set of purposes only, and does not allow for difference in collective organization. While
the goals or projected promise of modular robots are in search-and-rescue, planetary
exploration, industrial plant inspection and repair, this paper proposes a modular robotic
system that is architectural in size and scope (Figure 1).
Fig. 2

The current field of modular robots yields modules of generally a few inches in size. Necessary
restrictions in size are determined by available electro-mechanical parts, and fitness of uses.

This paper examines the precedent of the snake-like collection of room-sized modules and
how this system propels itself. It also looks at how a snake-like assembly can fold into compact
clusters. This examination is based on a scalar shift in the modules that allows for human
occupation so that buildings can be re-considered as mobile dwellings, as well as smaller units
that can cluster so that conjoined, larger spaces may be produced as needed.
Fig. 3

3.2 CURRENT WORK

3.2.1 GAIT CONTROL TABLES

A Gait Control Table is a matrix of values used to discretely parameterize the motion of a group
of Modular Robots. This method of control requires that each module have just one adjustable
parameter (one degree of freedom). This paper explores two types of module: for the first,
the adjustable parameter is a joint angle (see Section 4.1), and for the second the adjustable
parameter is the length of a truss (see Section 4.2). Each column corresponds to one module in
the group, and each row states the parameter values for a given step in the gait. Using variable
joint angle modules, reading an entire column gives a sequence of angles for that joint in the
corresponding module. A diagonal pattern of numbers can be observed in the Traveling Arc Gait
Control Table in the Appendix. This diagonal pattern results in the traveling wave (the wave that
propagates through the chain of modules, (Figures 2, 3).

3.2.2 FOLDING

In addition to moving in the vertical plane, modules can also move in the horizontal plane. In this
case, a chain of modules can bend to form a courtyard by placing several 90 degree horizontal
bends at even intervals in the chain, similar to Aldo van Eyck’s Municapal Orphanage (Figure 4).
Conversely, the modules could form a double row of rooms by having a 180 degree bend in the
middle module of the chain (Figure 5).

4 M o bi l i ty And C o nfi g urabi l i ty

Smithson’s defining work in a new modular architecture had another unique determinant in
movement and time: “For architects mobility has several connotations: in terms of movement
it signifies the shift from 2.5 miles per hour to 60,100, or 500 miles per hour. In terms of
time it means the appreciation of a fourth dimension i.e., change on a short time cycle. In
terms of economy, it means rapid mass-distribution, consonant with the potentialities of mass-
Figure 1. Example of search and rescue Modular
production and mass-consumption.”
Architecture

Figure 2. Wave gait To elaborate on her division of stem-to-cluster, she lays out the rule that it is mobility that is the
Figure 3. Two modules joined at docking plate metric: “These are the factors which vary from year to year and from place to place; and, if taken

352

acadia 2011 _proceedings integration through computation

as determinants of a scheme can give it organization and identity of a higher order than that
obtained solely from plastic arrangement. Stem is conditioned by mobility. Its dimensions are
given not in measures of length but of speed: 2.5 miles per hour and 60 miles per hour. It can
provide the link between these measures of speed as it can between the measures of validity
(the 25 and 5-year cycles).”

Smithson is defining a means by which this new mat-urbanism has validity in its part-to-whole
relationships. Another relationship however is in a changeability and reconfigurability that can
be realized. Given the nature of the movement and reconfiguration of this proposed system,
it is necessary to involve a discussion of Modular Robotics. More specifically, two robotic
prototypes will be introduced.

4.1 POLY-HOUSE1: FIXED GEOMETRY

This prototype consists of fixed geometry modules. Each P-HOUSE1 module (Figure 6) is
identical in size and material, consisting of a self supporting box-like light frame with a motor
actuated joint on one end and a docking plate on the other. One actuated degree of freedom
is permitted per module so that the only movement is rotation from a hinge. Motor actuation is
controlled by an on-board computer, and each motor side can dock with the docking plate of
the next module.

Fig. 4
Motion is achieved by some modules lifting neighboring modules such that the shape of the
chain of modules forms an undulating wave pattern in the vertical plane. This wave is propagated
down the chain to form a traveling wave that results in locomotion. Computers in each module
coordinate with other modules to determine which angles the motors should drive to in order to
form the appropriate traveling wave motion.

Whereas modular robots on the inch scale are dense with motors, batteries, electronics,
Fig. 5
sensors, etc., modular architecture in this sense needs to have a ‘payload’ for the robot
– a living space. Thus the primary engineering challenge is incorporating sufficiently large
actuators and structure to lift the modules while leaving enough room inside the modules
to be reasonable for occupancy.

Each module consists of four elements: a rectangular living space, a rotational motor, a
transmission actuator appended to the outside of this, and a docking plate on the opposite
side from the motor. When the docking plate mates with the motor side of another module,
both physical and electrical connections are formed. This connection can be made manually,
or autonomously as have been demonstrated in the smaller scale modular robots. Note that
autonomous docking would require more sophisticated sensing, computation, and consume
more space in the module.

Initial estimates are that an aluminum frame structure with approximately 200 hp engines in
each module would be capable of moving at walking speeds. Using this power source and
sequentially undulating the joints between a positive and negative 12 degree rotation, the
system would move at about 3mph. Increasing engine power or rotating to larger joint angles
would result in faster locomotion if desired.

If the modules are arranged with their joint axes aligned and perpendicular to both gravity
and the direction of the chain, the motion of the modules is in the vertical plane and the
locomotion described can be achieved. If the system includes some modules with their motor
axes normal to gravity (Figure 7) the global shape can also form lateral bends in the structure,
steer the system or to form a building of a courtyard typology. This complicates the motion,
in that the actuators must now lift more modules with a larger lever arm. As a result, the
torque requirements increase by a factor of roughly 4. The most complete and versatile version
of a P-HOUSE1 alternates between these two types of modules to maximize resolution (by
maximizing elements) in both dimensions.

4.2 POLY-HOUSE2: VARIABLE GEOMETRY

Fi gure 4. M u n ic ipa l Orph a n a g e , Am s t e r d a m ,

This prototype is a modular organization where the individual modules are dynamic and A l d o v a n E y c k - Top: plan - Bottom: model
changing in shape. This requires that certain members or structural chords of the unit frames Figure 5. Conceptual model of modules folding into
extend and contract. The base number of edges required for an irregular box (twelve), are larger aggregate space

353

kinetics , matter and computation

 Fig. 6

Fig. 7

maintained. However, the extensibility of its edges results in a greater number of hinges and
moving joints at their connections.

The basic assembly of this P-HOUSE2 is a six-sided frame in which the changing
dimensions of its members have direct influence over its modular form. Unlike the
P-HOUSE1 which has a motor external to the living space, the actuators are embedded
into the frame of the structure. As such, the connection plates at both ends that allow
for joining with other modules, are integrated within the living space as 2 of its 6 sides.
Note also that doors can be incorporated into the connection plates such that residents
may traverse from room to connected room without exiting the complex.

The two connection plates can form two rigid sides of the living space. However, the
rest of the shape is formed by the 8 edges connecting the opposing four corners
of each connection plate through free running spherical joints at each corner. This
structure has two degrees of freedom and would collapse without additional struts. Two
diagonal struts are added to each face to maintain the structural integrity. Three of these
8 members can change their length resulting in a change in a relative angle between
connection plates. This form is similar to a simplified version of what is called a Variable
Geometr y Truss (VGT). It is different in that VGT’s typically have more actuated elements
(Inoue et al. 2003).

Ever y module has an on-board computer that controls the actuation of each strut for
both the contraction/expansion of the structural frame as well as coordination between
modules to achieve locomotion or changing shape. These modules can achieve bending
angles in both the vertical and horizontal plane, either independently or in combination.

Rather than having one motor and a large transmission, this system has the actuation
distributed into the frame. Smaller motors such as a 5KW electric motor with high force
to torque ratio lead screws can be used that is much more space efficient. However,
the cost of this reduction in size is roughly an order of magnitude reduction in speed.

4.3 CLADDING: FLEXIBLE MEMBRANE

Figure 6. Poly-HOUSE1: Fixed geometry showing
frame and docking plates The prototype of modular organization necessitates a flexible modular enclosure system.
Figure 7. Poly-HOUSE1: fixed geometry Through implementation of triangulated concertina geometr y, the switch between planar

354

acadia 2011 _proceedings integration through computation

Fig. 11
structural systems and variable-angle module-to-module connections can be negotiated.
The concertina unit is defined by a triangulated three-division vertical panel with an Fig. 8
adjacent two-division panel. Each panel contains two mirrored triangular faces. This
overall unit connects to neighboring units through coincident edges. This unit can then
be applied as a modular system. The raising and lowering of panel division points allows
the overall triangulated pattern to var y its translation across the larger aggregate.

Such a geometrical system paired with lightweight construction techniques allows for
long spanning modules, surface expansion, as well as a connectivity of enclosure.
Planar units constructed out of traditional membrane types allow for ease of manufacture
through conventional fabric joining techniques. Interchangeable membrane materials
such as canvas, polyethylene films, and other breathable membranes suggest the Fig. 9

variability of enclosure situations that can be implemented to fit programmatic needs of

the module. Lightweight bent metal frames allow membrane supports to be attached
to the module frame without additional structure. This would allow for larger aggregate
spaces to be produced when modules fold from a snake-like formation to a larger block.

Further architectural characteristics arise through the variability of façade articulation.

The triangulated distributions of panels allow shifts of pattern and movement, breaking
down the imbedded banality and repetition of the existing modular architecture
tradition. The potential variation of material further suggests a range of experiential
qualities that result from shifts in density of folds. In this case, the transformational
effects of locomotion promote an aesthetic indeterminacy through layering and
s h i f t i n g d e n s i t y o f p a t t e r n ( F i g u re s 8 - 1 5 ) .

4.4 OCCUPANCY: ENFILADE OR SINGLE-LOADED CORRIDOR

Each module is of sufficient depth at 3m: a single loaded corridor can thus be provided Fig. 10
so that privacy among occupants can be maintained. In a courtyard configuration, the
typology of open space and loggia is maintained.

In a maximization of interior space, the modules may be kept open at 3m by 6m and

procession between modules is through the architectural enfilade. This option has a
tradeoff in separation and privacy, but gains in greater assembly when modules are
packed into a larger block. Furthermore, the uninterrupted space allows for flexibility of
layout and indeterminacy of program of mat-urbanism.

5 Con c lu sion s a n d Fu tu re Wor k

Modular reconfigurable robots promise to be extremely versatile, and low cost per
repetition. These systems are made up of many self-similar simple modules that can be
arranged in many different configurations.

As put forward by Smithson, mat-urbanism in the format of modular architecture has

high potential of changing occupancy in two primar y ways. The first effect concerns
Figure 8. Poly-HOUSE2: showing walking gait of
shared public and private space, and the aggregation of rooms: one modular system
flexible modules
can house many types of functions and dynamically adjust to its changing needs.
Figure 9. Poly-HOUSE2: variable geometry
Modules can continually chain together to become larger settlements, or can separate
for solitar y requirements. Secondly, the idea of architecture and place (locus) become Figure 10. Example of PolyHouse1 cladding

untethered so that ideas of urban or rural dwelling, site and context, are no longer static. Figure 11. Definition of folded cladding

355

kinetics , matter and computation

 Fig. 13

Fig. 14

Communities can expand or contract, and likewise, a modular family residence can grow
or shrink by the addition or removal of modules, as well as move en masse. This may
result in minimizing supply chain issues of remote har vesting and deliver y, and that local
resources remain renewable, as communities can move in nomadic pursuits.

For Modular Robotics, this current research can expand into different scales of modules.
Differing geometries, such as quadrilateral polyhedral for vertical stacking into towers,
can also be explored.

Currently larger-scale models have been designed and motion has been validated in
simulation software. Future work is bound in the fabrication of full-scale prototypes. The
development of modular robots at a smaller scale prove the architectural theor y, but
validation of linear movement, folding into packs of larger blocks, require field testing.
Current developments in electro-mechanical technology would provide reductions in
energy consumption as well as long-term energy har vesting and storage.

R e fe re n c e s

Allen, S. 2001. Mat urbanism: The thick 2-D. In. Case: Le Corbusier’s Venice Hospital and the Mat
Building Revival, ed. S. Sarkis, P. Allard, and T. Hyde, 118-26. New York: Prestel.

Chirikjian, G. S. and J. W. Burdick. 1995. The kinematics of hyper-redundant robot locomotion. IEEE
Transactions on Robotics and Automation 11(6), 781-793.

Fukushima, E. F. and S. Hirose. 1996. Efficient steering control formulation for the articulated body
mobile robot kr-ii. Autonomous Robots 3(1), 7-18.

Inoue, F., K. Kurita, Y. Utsumi, and N. Furuya. 2003. Application of adaptive structure and control by
variable geometry truss, Proc. of The CIB 2003 International Conference on Smart and Sustainable
Built Environment, 5.

Sarkis, H. 2001. Introduction, in Sarkis, S., P. Allard, and T. Hyde (eds). Case: Le Corbusier’s Venice

Figure 13. Rendering

Hospital and the Mat Building Revival. New York: Prestel, pp.12-17.

Figure 14. Rendering

356

acadia 2011 _proceedings integration through computation

Fig. 15

Smithson, A. 2001. How to recognize and read mat building, in Sarkis, S., Allard, P. and Hyde, T. (eds)
Case: Le Corbusier’s Venice Hospital and the Mat Building Revival. New York: Prestel, pp.90-103.

Yim, M., D. G. Duff, and K. D. Roufas. 2000. Poly-bot: A modular reconfigurable robot. In: International
Conference on Robotics and Automation. IEEE. San Francisco, California, USA. pp. 514-520.

Appendix

Gait Control Tables

Figure 15. Rendering

357

kinetics , matter and computation