Independent joint dynamic sliding mode control of a humanoid robot arm

Abstract

Since a rigid humanoid robot arm is kinematically redundant and composed of seven degrees of freedom at least to execute complex and dexterous motions, how to construct a fast and precise control system becomes an important issue. To avoid high complexity of the high-order dynamic model considering coupling effects and payload variations typically inherited from common robotic systems, a robust dynamic sliding mode control law design based on an independent joint model, tremendously reducing the complexity of the dependable design process, is proposed to control a Multi-DOF humanoid robot arm with parameter variations, external disturbances and payload variations. The global asymptotic stability of the controller is guaranteed by the Lyapunov approach, and the input chattering phenomenon is handled by using the augmented system. Through dynamic simulation and practical experiments of the NTU Humanoid Robot Arm II, the results show that the proposed controller is better than other kinds of control approaches.