ABSTRACT
To achieve desirable biomimetic motion, actuators must be able to reproduce the important features of natural
muscle such as power, stress, strain, speed of response, efficiency, and controllability. It is a mistake, however, to
consider muscle as only an energy output device. Muscle is multifunctional. In locomotion, muscle often acts as an
energy absorber, variable-stiffness suspension element, or position sensor, for example. Electroactive polymer
technologies based on the electric-field-induced deformation of polymer dielectrics with compliant electrodes are
particularly promising because they have demonstrated high strains and energy densities. Testing with experimental
biological techniques and apparatus has confirmed that these “dielectric elastomer” artificial muscles can indeed
reproduce several of the important characteristics of natural muscle. Several different artificial muscle actuator
configurations have been tested, including flat actuators and tubular rolls. Rolls have been shown to act as structural
elements and to incorporate position sensing. Biomimetic robot applications have been explored that exploit the
muscle-like capabilities of the dielectric elastomer actuators, including serpentine manipulators, insect-like flappingwing
mechanisms, and insect-like walking robots.
Keywords: biomimetic, artificial muscle, electroctive polymers, dielectric elastomers, robots, walking
1. INTRODUCTION
It has long been a goal of roboticists to achieve lifelike motion. This goal is perhaps partly motivated by
anthropomorphism, yet there are powerful technical motivations to developing lifelike robots. The entertainment
industry, with its need for humanoid robots and robotic creatures, is one major application area that can benefit from
lifelike robots. Beyond mimicking the appearance of natural organisms, there are a number of strong performance
motivations for achieving lifelike performance, as opposed to lifelike appearance. Natural creatures put man-made
robots and devices to shame when it comes to navigation of obstacles, speed over rough terrain, agility, and, in many
cases, power or energy output per unit weight in accomplishing certain tasks.
One immediate, and probably the most significant, obstacle in achieving lifelike appearance or performance is the
lack of a commercial actuator technology that can truly mimic natural muscle even at its most basic performance.
Table 1 shows a qualitative comparison between natural muscle and various commercial actuator technologies. If we
look closely at quantitative performance, we find that although natural muscle is not the best in many individual
categories of performance, it is good in virtually all measures of performance. This leads to the observation that
conventional technologies fail to achieve lifelike motion not because they cannot match or exceed natural muscle
performance in any given performance measure, but rather they fall short because they do not equal natural muscle
in overall performance.

Download full report
http://ciber.berkeley.edu/twiki/pub/CIBE...c_SPIE.pdf