Background

Traditional robotic actuation is done via electric motors or pneumatics/hydraulics. 

Electromagnetic step motors offer precision, use a convenient form of power, and have some capability for miniaturization, making them the usual choice for small robots and prosthetics. However, these motors are electromagnetic (EM) motors, which require a strong magnetic field generated either by strong permanent magnets or solenoids running large currents. Conventional EM motors often choose the latter path and require significant power to operate, while generating excess heat. 

Pneumatic systems provide more force in large systems, e.g. construction vehicles, industrial assembly lines, the US Army’s Mule walking robot, etc., but they require compressors, can spring leaks, and output less force when scaled down for use in compact systems. Furthermore, complex fluid motions are difficult to achieve by pneumatics because pressure is typically either on or off, producing jerky choppy motion that may be acceptable in an industrial robot but impractical in exoskeletons, prosthetics, etc. 

Due to these limitations, a wide range of applications requiring actuation, such as exoskeletal locomotion, walking robots, biomimetic underwater propulsion, prosthetics, medical servo-assists, and small-scale biomimetic robots, look to different actuation systems as a potential solution, including artificial muscles. Artificial muscles can be organized in several large groups: piezoelectric, pneumatic artificial muscles (PAM), thermal actuators, and electroactive polymers (EAP).

Piezoelectric actuators offer large forces in small devices at low voltages but the range of motion is very small. Devices are often stacked to mitigate that drawback. For example, such stacks are successfully used in the beam control circuitry of atomic force microscopes. However, the cost of individual devices and manufacturing difficulties severely limit the size of practical stacks, with the resulting overall elongation still being too small for typical artificial muscle applications. 

PAMs, e.g. McKibben muscles, cloth muscles, and RIPAs, employ a flexible bladder structure enmeshed in braided, crisscrossed, or wound fibers. As the bladder is filled with air, it deforms and displaces the fibers, outputting force. While possessing advantages in compactness and force output compared to piston system, PAMs use the same basic principles as hydraulics/pneumatics and thus suffer from the same basic limitations in unfavorable scaling and control issues.

Thermal actuation has also been proposed, e.g. with anisotropic materials that curl up with a temperature change, producing torsional artificial muscles and shape-memory-alloys muscles (SMA). Thermal expansion and contraction can generate high forces, but heat transfer severely limits the thermal actuators’ response speed and cycling frequency. As a result, such actuators are not practical for most propulsion applications.

EAPs change shape under the influence of an applied electric field. They are considered closest to the biological muscles among all the above-mentioned approaches. They avoid the use of magnetic fields and thus avoid the concomitant limitations. However, EAP actuators are typically complicated heterogeneous materials that are difficult to fabricate and suffer from low reproducibility, very low efficiency, and low durability. As a result, they have proven very difficult to manufacture to the standards and at the scale and price-point required by practical applications.

Dielectric Elastomer Actuators (DEA) are a particular subclass of EAP, wherein the actuation is a result of the deformation of a polymer (elastomer) slab under the electrostatic force between the charges built on the slab’s surfaces under applied voltage. That force is small macroscopically, but it scales as the inverse square of the separation between the plates. Hence, shrinking the devices to the microscale would gain a disproportionate increase in force. Arraying such devices in 3D should increase both force output and motion distance. However, manufacturing such arrays from traditional materials (e.g. metal electrodes and polymer dielectrics) by traditional manufacturing means (e.g. photolithography) to the necessary scale is impractical for reasons similar to the difficulties experienced with the piezoelectric and EAP approaches.

Based on the above analysis, the need for practical artificial muscles remains unmet.