Physics-Based Finite Element Simulation of the Dynamics of Soft Robots

Soft robots grew to prominence in large part because they promised a new and exciting means for engineers to develop robots that are both highly adaptable and safe for direct human interaction. However, despite showing substantial promise in this area, soft robots have not yet seen widespread adoption. Two major factors that have prevented the development of soft robots are the fundamental challenges of both the modeling and control of soft structures. While traditional robots enjoy a myriad of theoretical and computational tools for modeling and control, the options for soft robots are far more limited. In this work, we introduce a physics-based finite element simulation platform, Kraken, that can be used to accurately model the dynamic and oscillatory motions of soft robots. After a brief theoretical introduction to hyperelastic modeling and the finite element method, we show the utility of our approach by simulating the oscillations of a 1D hyperelastic actuator, the dynamics of a 2D hyperelastic pendulum, and a 3D spherical hyperelastic pendulum. We then demonstrate the accuracy of our approach by presenting the agreement of the simulated results with those obtained via physical experiments for three materials with different hyperelastic properties, with percent errors as low as 1%. Taken together, these results demonstrate that the aforementioned simulation platform is a critical step towards the fast and accurate simulation, prototyping, and control of soft robots.