Computational Mechanics for the Design, Actuation and Control of Soft Robots

Abstract

The compliance, resilience and adaptability of biological systems have sparked the emergence of a nascent class of robots, that are soft, inexpensive, easy to fabricate and capable of safely interacting with their environment. Soft robotic research is constantly extending the capabilities of soft robots by utilizing their compliance and nonlinear deformation to enable a variety of innovative applications. However, strategies for the design, actuation and control of soft robots are predominantly dictated by experimental prototyping with limited influence from simple analytical models and computational tools. Computational mechanics have historically supplemented theory and experiments in all disciplines of physics and engineering by complementing any analytical solutions, providing insight to complex nonlinear problems and minimizing the number of experiments required. Additionally, the rapid growth of available computational resources combined with the increased efficiency and robustness of computational methods, have enabled the solution of problems with unprecedented complexity. In this dissertation, I embrace computational mechanics and utilize numerical methods, finite-element models and optimization tools to instruct the design but also simplify the actuation and control of soft robots by capitalizing on their nonlinearities. In particular, I develop a unified finite-element based numerical tool which enables the simulation of laminar and fibrous media undergoing a jamming transition and facilitates the design of jamming-enabled soft robots. Furthermore, through a combination of finite-element simulations, numerical models and optimization tools, I demonstrate that viscous flow can dramatically simplify the actuation of fluidic soft robots. Finally, I develop models to study the bidirectional propagation of transition waves in arrays of universally bistable shells with nonzero Gaussian curvature, proposing an alternative strategy for a purely mechanical control of soft robots. Altogether, the numerical models, optimization tools and finite element simulations presented in this work, demonstrate the potential of computational mechanics in accelerating the design and experimental prototyping of soft robots.