466 日 本 ロ ボ ッ ト学 会 誌Vol.24No.4,pp.
466∼470,2006
解 説
Electroactive Polymer
Seiki Chiba*, Scott
1. EPAM Background
Many types of candidate electroactive materials have
been or currently are under investigation, including
single-crystal piezoelectric ceramics and carbon nan-
otubes. Electroactive polymers (EAPs), such as those
developed by SRI, are of particular interest because of
the low cost of the materials and the ability of polymers
to be tailored to particular applications. Piezoelectric
polymers such as PVDF have been known for many
years. This material has found some non-robotic appli-
cations, but for robots PVDF has a relatively low power
and energy density. More recently, many other types
of polymers have been investigated. Electrochemically
actuated conducting polymers and gels, sometimes re-
ferred to as ionic polymers, use chemicalchanges driven
by low voltages to actuate. Examples include conduct-
ing polymers and IPMC. These polymers are attractive
because of their low voltage operation and good actua-
tion pressures, but are limited for robotic applications
because of their relatively slowspeeds and lowefficiency.
Another category of electroactive polymers are some-
times referred to as electronic EAPs. This category in-
cludes EPAM as described below, and also electrostric-
tive polymers[1], which change dielectric constant as
part of the actuation mechanism. Electrostrictive poly-
mers have comparable performance to many types of
EPAM, but are typically stiffer and have lower strains.
In general, electronic EAPs require much higher volt-
ages than ionic EAPs, but they can be fast, efficient,
and have high energy output.
Starting in the early 1990s, SRI invented a new type
of electroactive polymer, knownvariously as EPAM, di-
原 稿 受 付2006年1月10日
キ ー ワ ー ド: Dielectric Elastomer, Electroactive
tificial Muscle,Robots,EPAM
*333 Ravenswood Ave
.,Menlo Park,CA 94025
JRSJ Vol.24 No.4
Artificial Muscle
Stanford*, Ron Pelrine*, Roy Kornbluh* and Harsha Prahlad*
*SRI International
electric elastomers, or electroelastomers. It has been
well known for many years that the electric field pres-
sure from free charges on the surface of all insulating
materials induces stress (Maxwell's stress) that strains
the material. Previously, Maxwell stress was regarded
as a nuisance effect on piezoelectrics,but when applied
to softer polymers with high breakdown voltages, the
mechanical output can be substantial. The basic ele-
ment of EPAM actuators is a dielectric polymer film,
typically 10 to 200 [µm] thick, that is coated on each
side with a compliant electrode material such as carbon-
impregnated elastomer. Normally the polymer is an
elastomer for highest strains, but non-elastomer poly-
mers can also produce significant actuation. When a
voltage is applied across the two electrodes, the electro-
static forcescompress and stretch the film. As shownin
Fig. 1, the compression of the film thickness brings op-
posite charges closer together, whereas planar stretch-
ing of the film spreads out or separates similar charges.
Both changes convert electrical energy to mechanical
energy and provide the actuation mechanism.
Strains as high as 380% have been measured, with
actuation pressures up to 8 {MPa]. The available lit-
erature indicates that the actuated strains of dielectric
elastomers are greater than that of any other known
high-speed electrically actuated material (i.e., material
actuated at a bandwidth above 100[Hz]). Dielectric
elastomers also have other desirable material properties
Polymer,Ar-
Fig.1 Applied voltage causes dielectric elastomer EPAM
to decrease in thickness and expand in area
USA
38 May,2006
 Electroactive Polymer Artificial Muscle 467

•E High efficiency in power conversion: Many future

robot applications require mobility with onboard

power supplies, requiring high efficiency actuation.

•E Multifunctionality: The ability to use a single device

for multiple functions such as multiple degrees of free-

dom (DOF) in a single actuator or combining struc-

tural and actuator functions or combining actuation

and sensing.

Fig. 2 Performance of dielectric elastomer is similar to that 3. First Generation of EPAM-Enabled Robots
of natural muscle
In seeking to create robots that embody the ad-

such as good actuation pressures and high theoretical vantages discussed above, we have built many first-

efficiencies(80 to 90%)because of their low viscoelastic generation prototypes. Some of the most striking exam-

losses and low electrical leakage. In addition, the en- ples are biologically inspired robots that use cylindrical

ergy density of dielectric elastomer EPAM has reached •g spring roll•h actuators made from dielectric elastomers.

3.4 [J/g], about 21 times that of single-crystal piezo- These robots are shown in Fig. 3. A spring roll actuator

electrics and more than two orders of magnitude greater consists of electroded elastomer film wrapped around a

than that of most commercial actuator materials [2] . As spring. When an applied voltage actuates the film, the

can be seen in Fig. 2, dielectric elastomers not only out- spring extends axially, creating a 1-DOF spring roll. By

perform existing actuator technologies in various areas patterning multiple electrodes on the film, one can also

but also are similar to natural muscle in that they fill make an actuator that can bend in multiple directions,

the •gactuator gap•h between other actuation technolo- creating 2- or 3-DOF spring rolls.

gies. Flex [3], the largest of the robots at 470 [g] mass and

36 [cm] length, was the first self-contained (i.e., battery

2. Desired Properties of New Robot Actuators
powered) walking robot powered by electroactive poly-

Some of the key desired features for new robotic ac- mers (Fig. 3 (a)). A second version of this robot used

tuators (whether polymer or other types) include the two 1-DOF spring rolls for each of its six legs, one ac-

following: tuator to lift and lower the leg, and the other actuator

•E Totally quiet operation: This is important for noise- to move it forward and back [4] . Like the two robots

sensitive applications such as in hospitals, home described below (and most insects), Flex walks with a

robots, and stealth for military or law enforcement •g dual tripod•h gait in which the left front, left rear, and

robots. center right legs move up, down, forward, and back to-

•E Variety of shapes and sizes, from cylindrical, like gether. Flex can move at speeds greater than 12 [cm/s],

many skeletal muscles, to flat, such as proof-of- actuating its legs at over 10 [Hz].

principle EPAM devices that can travel under doors.•E Skitter [5] was designed to demonstrate the ability of

High power density and high peak power density, to EPAM to act not only as an actuator but also as the

enable rapid dynamic motions such as those used in structure of a robot as a first step in multifunctionality

hopping, jumping, or running to surmount obstacles (Fig. 3 (b)) : This particular robot is based loosely on
or for high speed maneuvers. the Sprawlita robot [6], which uses pneumatic actuators

•EInherent springlike compliance, to enable energy to in a similar fashion. Each of the six legs points a sin-

be stored for efficient locomotion and adaptation to gle 1-DOF actuator down and slightly backward, and is

uneven terrain or cluttered workspaces. Natural crea- attached to the body on a compliant joint. In spite of

tures use the compliance of muscles and tendons to its simple design, Skitter has demonstrated a speed of

create oscillatory systems that locomote efficiently. 6.8 [cm/s].

Unlike highly geared motors, the joints should be back MERbot [7], which gets its name from the Multifunc-

drivable to adapt to uneven terrain or easily recover tional Electroelastomer Rolls (MERs) that form its legs,

from collisions. is the next logical step, using 2-DOF rolls as flexible legs

日本 ロ ボ ッ ト学 会 誌24巻4号 39 2006年5月
468 Seiki Chiba Scott Stanford Ron Pelrine Roy Kornbluh Harsha Prahlad

Fig. 3 Biologically inspired EPAM-powered robots (Movies are available at

http://www.erg.sri.com/video.html)

Fig. 4 Proof-of-principle jumping EPAM device

in lieu of discrete hip or knee joints (see Fig. 3 (c)). The

MERs yield a very light, simple robot that still has a
high degree of dexterity. MERbot has traveled faster Fig. 5 Conceptual design of an EPAM-based snake robot
than 14 [cm/s].
Jumping robots are also attractive with EPAM, and
this application illustrates the advantages of high en-
ergy direct drive actuation. Fig. 4 shows an early jump-
ing EPAM platform [7]. Note from Fig. 4 the potential
simplicity of a jumping robot EPAM design. This de-
sign uses only three EPAM diaphragm actuators. Elec-
tric motors have difficulty generating the high peak
power density necessary for jumping starting from a
stopped state, so electric motor jumping drives typi-
cally require gearboxes and energy storage in a spring
with a sudden release mechanism. Fig. 6 Conceptual design of a raylike swimming robot

4. Future Generations of EPAM-Enabled

Robots
One design under consideration is a snake robot,
which could be manufactured by concatenating several
3-DOF rolls (see Fig. 5). SRI has demonstrated proof-
of-concept snake segments. Snake robots are challeng-
ing from both a mechanical and a control perspective.
For snakes and tentacle-type robots, besides offering
tally low, with estimates of 0.05 [W/g] compared with
0.1 to 0.2 [W/g] for natural muscle. Electric motors are
also inefficientin the start-stop mode needed for snake
mobility. Other efforts to reproduce musclelikeactua-
tion in snakelike robots have used shape-memory alloy
actuation [10],but again these snake robots are slow and
inefficient.

performance advantages, the simplicity and multifunc- EPAM is also uniquely qualified to produce finlike

tionality of EPAM is a major advantage because of the motion, including the rajiform motion used by skates

number of DOF required. and rays. A swimming •gRaybot•h (Fig. 6) would be

Electric motors are typically used for snake robots [8] highly maneuverable and able to rotate in place or travel
backward. Its flat shape would be less susceptible to
[9]. However,because of the need for gear transmissions
to reach desired torques, overall power density is typi- turbulence and would also afford more space for sensors

JRSJ Vol.24 No.4 40 May, 2006

 Electroactive Polymer Artificial Muscle 469

(www.artificialmuscle.com) is making significant strides

in the commercialization of EPAM in the areas of lower

voltage, longer lifetime, and greater environmental op-

erating ranges. This base technology development en-

sures future EPAM robot designs will enable more rad-

ical robot designs and functionality than might be en-

visioned today.

References

[1] Q. Zhang, V. Bharti and X. Zhao: •gGiant elect restrict ion and

relaxer ferroelectric behavior in electron-irradiated poly (vinyli-

dene fluoride-trifluoroethylene) copolymer,•h Science, vol.280,

Fig. 7 Conceptual design of a flapping wing robot
pp.2101-2104,1998.

[2] R. Pelrine, R. Kornbluh, Q. Pei and J.P. Joseph: •gHigh-speed

electrically actuated elastomers with over 100% strain,•h Sci-

or other payloads. Traditional actuation methods are ence, vol.287, no.5454, pp.836-839, 2000.

[3] J. Eckerle, S.E. Stanford, J.P. Marlow, R.H. Schmidt, S. Oh,

clearly inadequate for such motion. A previous effort
T.P. Low and S.V. Shastri: •gA biologically inspired hexapedal

to reproduce musclelike actuation in a rajiform robot robot using field-effect electroactive elastomer artificial mus-

cles,•h Proc. SPIE 8th Annual Symposium on Smart Structures

(Robo-Ray, the product of a student project at the Uni-
and Materials 2001: Industrial and Commercial Applications of
versity of British Columbia) used a shape memory alloy,
Smart Structures Technologies, vol.4332, 2001.

but this material was found to be •ginefficient and diffi- [4] R. Pelrine, R. Kornbluh, Q. Pei, S. Stanford, S. Oh, J. Eckerle,

R. Full, M. Rosenthal and K. Meijer: •gDielectric elastomer ar-

cult to control•h [11].
tificial muscle actuators: Toward bioinimetic motion,•h Proc.

EPAM could also enable other biomimetic propulsion SPIE, Smart Structures and Materials 2002: Electroactive

Polymer Actuators and Devices (EAPAD), vol.4695, 2002.

technologies such as peristaltic movement, or jet propul-
[5] Q. Pei, R.E. Pelrine, S.E. Stanford, R.D. Kornbluh,
sion such as that used by jellyfish and squid. The lat- M. Rosenthal, K. Meijer and R.J. Full: •gMultifunctional elec-

ter application would exploit the inherent compliance troelastomer rolls and their application for biomimetic robots,•h

Proc. SPIE, Smart Structures and Materials 2002: Industrial

of EPAM and the fact that it matches the mechanical and Commercial Applications of Smart Structures Technolo-

impedance and density of water. gies, vol.4698, 2002.

Dielectric elastomers are also attractive for flapping- [6] J.E.

P.K.
Clark,
Nahata,
J.G.
R.J. Full
Cham,
and M.R.
S.A. Bailey,
Cutkosky: •gBiomimetic
E.M. Froehlich,
de-

wing flight (see Fig. 7); in fact, various EPAM-powered sign and fabrication of a hexapedal running robot,•h Proc. IEEE

flapping mechanisms have been manufactured and Intl. Conf. on Robotics and Automation (ICRA), 2001.

[7] Q. Pei, M. Rosenthal, R.E. Pelrine, S.E. Stanford and

tested in the laboratory [12]. These flapping mecha- R.D. Kornbluh: •gMultifunctional elect roelastomer roll actu-

nisms have not yet produced the continuous power nec- ators and their application for biomimetic walking robots,•h

Proc. SPIE, Smart Structures and Materials 2003: Electroac-

essary for self-containedsustained flight, because of ex- tive Polymer Actuators and Devices (EAPAD), 2003.

cessive structural mass for the actuator (the EPAM [8] G.S. Chirikjian and J.W. Burdick: •gKinematically optimal

hyper-redundant manipulator configurations,•h IEEE Trans.

fraction was only about 10% in tests to date) but they Robotics and Automation, vol.11, pp.794-806, 1995.

do show the ability for a very simple structure that can [9] M. Yim, D. Duff and K. Roufas: •gPolyBot: A modular re-

use resonance to achieve high flapping amplitudes at configurable robot,•h Proc. IEEE Int. Conf. on Robotics and

Automation (ICRA), 2000.

high power densities. [10] C. Wilbur, W. Vorus, Y. Cao and S. Curie: •eA lamprey-based

undulatory vehicle,•f in Neurotechnology for Biomimetic Robots.

5. Summary and Conclusions eds. J. Ayers, J. Davis and A. Rudolph, MIT Press, 2002.

[11] H. Davis, R. Boileau, L. Fan and T. Moore: •gMechanization of

Electroactive polymer artificial muscles based on di- rajiform swimming motion: The making of Robo-Ray,•h student

project, University of British Columbia Engineering Physics

electric elastomers have been successfullyused in a di- Project Laboratory, January, 2002.

verse array of robotic applications. Several proof-of- [12] R.D. Kornbluh, R.E. Pelrine, Q. Pei, R. Heydt, S.E. Stanford,

S. Oh and J. Eckerle: •gElectroelastomers: Applications of di-

principle robots have been demonstrated. We antic- electric elastomer transducers for actuation, generation and

ipate continued advances in robots based on EPAM, smart structures,•h Proc. SPIE, Smart Structures and Mate-

rials 2002: Industrial and Commercial Applications of Smart

as it enables high performance in simple, low cost, Structures Technologies, vol.4698, pp.254-270, 2002.

multifunctional designs. Besides continuous improve-

ment by research organizations, Artificial Muscle, Inc.

日本 ロボ ッ ト学 会 誌24巻4号 41 2006年5月
470 Seiki Chiba Scott Stanford Ron Pelrine Roy Kornbluh Harsha Prahlad

Seiki Chiba Scott Stanford

Dr. Seiki Chiba (seiki.chiba@sri.com) is Mr. Scott Stanford is a Research Engi-
Executive director for Advanced R&D neer in the Mobile Robots and Transducers
Project Development. He has been super- (MRAT) program in the Engineering and
vising advanced R&D programs including Systems Division at SRI International. Mr.
Japanese Government projects. He is the Scott Stanford has led and participated in
author or coauthor of more than 40 publi- numerous mobile robots projects including
cations in the various areas including artificial muscle actua- MERbot, a walking robot using a bending EPAM spring
tors, hydrogen safety, and high temperature membranes for rolls as each of its six legs, and FLEX, a six legged, 12 DOF
hydrogen production. He received his Ph.D. in Metallurgy robot. Mr. Stanford received his B.S. and M.S. in mechani-
& Material Science at the University of Wales (Britain). cal engineering from Stanford University.

Ron Pelrine Roy Kornbluh

Dr. Ron Pelrine is Director of ESD's Mobile Mr. Roy Kornbluh, Senior Research Engi-
Robotics and Transducers Program. Dr. neer, has 15 years of experience working
Pelrine has been working with electroactive with EPAM actuators. Mr. Kornbluh has
polymer technology for the past fourteen served on the International Program Com-
years and has led efforts to apply the tech- mittee of the Electroactive Polymers and
nology for a variety of applications. Dr. Devices Conference of SPIE, the Actuators
Pelrine has a B.S. in physics from the Massachusetts Insti- Conferences, and the International Association of Science
tute of Technology, an M.S. in physics from the University and Technology for Development Robotics and Applications
of Washington at Seattle, and a Ph.D. in mechanical engi- Conference. He has an M.S. in mechanical engineering from
neering from the University of Texas at Austin. the Massachusetts Institute of Technology and a B.S. in me-
chanical engineering from Cornell University.
Harsha Prahlad
Dr. Harsha Prahlad, Research Engineer,
worked with a variety of smart materials in-
cluding EPAM, shape memory alloys, and
piezoceramic materials. He has an M.S.
and Ph.D. in aerospace engineering from
the University of Maryland, and a B.E. in
mechanical engineering from Regional Engineering College
in Trichy, India.

JRSJ Vol.24 No.4 42 May, 2006