Computational Modeling and Design of Soft-Robotic Pneumatic Fingers for Functionality and Durability

Abstract

Soft robotic actuators, such as pneumatic bending bellow actuators, have shown promise as grippers for robust, adaptive, gentle grasping. The effects of select geometric parameters on bellow actuator pressure-curvature performance have been explored by others using finite element (FE) models. We expand on this work by implementing a highly automated finite- element-based tool for the rapid and rigorous exploration of the bellow actuator design space with a focus on multi-objective functionality. Capable of parameter sweeps, grid searches, or evolutionary (CMA-ES) searches, our tool enables the investigation of fourteen geometric parameters relative to six objective metrics: maximum strain, curvature, bending angle, actuation pressure, actuation volume, and actuation energy. With our tool, we explored over 400 bellow actuator designs. First, we elucidated the effect of each of the geometric parameters on each of the objectives via comprehensive parameter sweeps. Next, we conducted a grid search to optimize actuator design for both durability and bandwidth. Then, we ran an evolutionary search across four parameters to maximize durability. We identified a bellow actuator design that exhibits a 60 % reduction in maximum strain and a 52 % reduction in actuation volume relative to the original design. Lastly, we demonstrated that our tool can be extended to the optimization of dexterity and durability for multi-DOF pneumatic actuators. We believe that computational, holistic, and multi-objective frameworks will enhance the design of the next generation of durable and functional soft robotic actuators.