Dielectric Elastomer Actuators as Artificial Muscles for Soft Robotic Applications

Abstract

Dielectric elastomer actuators (DEAs) are electro-mechanical transducers which can convert an electrical input into mechanical work. They have clear advantages (i.e., soft elastomers, direct electrical to mechanical conversion) which make them particularly promising for soft robotics applications. Unfortunately, most DEAs have severe limitations: low output forces, the need for a rigid frame, high voltages needed for actuation, slow response speed, low energy density, and challenging integration with other components. This thesis describes a material-focused approach to address these limitations, using unique robotic demonstrations as proof of the material and device capabilities. To ensure the final DEA is completely soft, I focused on strain stiffening elastomer materials, made from UV curable liquid precursors. These are combined with ultra-thin carbon nanotube (CNT) mats in a multilayering process which yields robust actuators, capable of bending, linear expansion, as well as contraction. The process gives control over each layer's thickness, meaning low voltage demonstrations are possible, such as bending at 300-600 V. Additionally, each electrode can be made from a specific ink, processed independently from the rest of the multilayer. Using thermally treated electrodes at low areal densities of CNTs, I built multilayers that can be charged to high electric fields (>200 V/micron), to do mechanical work at high energy density (10-20 J/kg). Using a multilayer bending beam, I demonstrated a crawling robot, capable of high speed locomotion (>1 body length / second). The crawling speed was set by the viscoelasticity of the elastomer, showing the strong dependence of robot performance on material properties. Additionally, the same bending beams were built into a bimorph used to power a swimming robot. By incorporating a pressure sensor and reconfiguring the robot to dive, I demonstrated the first ever autonomous DEA-powered robot. Lastly, using high energy density DEAs and a composite structure, I showed the first ever DEA-powered jumping robot. Overall, the multilayer platform allowed me to use material improvements (elastomers with lower viscoelastic losses, electrodes with higher conductivity, etc.) to build robots with unprecedented capabilities (high energy density, large bandwidth, impulsive behavior, etc.).