Modeling and control of coupled soft viscoelastic actuators based on sparse identification method

Abstract

Dielectric elastomer actuators (DEAs) are extensively employed as artificial muscles in bioinspired soft robotics due to their large voltage-induced deformations and muscle-like characteristics. Achieving complex, multiple degree of freedom(DOF) motions often requires coupling multiple DEAs. However, modeling and controlling coupled DEAs pose significant challenges due to their inherently nonlinear response, driven by factors such as rate-dependent viscoelasticity, design irregularities, and complex interactions among adjacent actuators. This study presents a comprehensive framework for the modeling and control of multiple coupled DEAs, leveraging a sparse identification approach to derive explicit governing equations that effectively describe the viscoelastic and coupling effects from experimental data. Using these identified equations, we design a model predictive controller (MPC) that enables precise trajectory tracking across a range of motion profiles. The proposed framework is validated through tracking experiments on DEAs with two DOFs, demonstrating its effectiveness and robustness. This approach offers a novel pathway for uncovering the underlying physics of coupled DEAs, with the potential to enhance the functional capabilities of DEA-driven soft robotic systems.