Sliding-mode control of a soft robot based on data-driven sparse identification

Abstract

Soft robots are increasingly finding their way into many applications, especially those involving manipulation of sensitive and delicate objects or interaction with humans. However, their high-compliance characteristics pose considerable challenges in obtaining low-complexity yet accurate dynamical models that are suitable for advanced feedback control. This paper proposes a framework for end-effector positioning of a soft robot. First, physics-informed sparse regression is used for deriving a nonlinear mathematical model of the robot dynamics. Then, a control scheme comprising a super-twisting sliding mode controller and a nonlinear input estimator is designed for the positioning of the robot end-effector. Conditions for uniform asymptotic stability of the closed-loop system are given. Finally, experimental tests carried on a real soft robot show the efficacy of the proposed design and its tracking accuracy.