Robust-tracking control for robot manipulator with deadzone and friction using backstepping and RFNN controller

This study deals with a robust non-smooth non-linearity compensation scheme for the direct-drive robot manipulator with an asymmetric deadzone, dynamic friction in joints and between the environmental contact space and end-effector and uncertainty. A model-free recurrent fuzzy neural network (RFNN) control system to approximate the ideal backstepping control law is designed to replace the traditional model-based adaptive controller, which requires information on the robots dynamics in advance. The simple dead-zone estimator and friction compensator based on the elasto-plastic friction model are developed in order to estimate unknown dead-zone width and friction parameters. The Lyapunov stability analysis yields the adaptive laws of the RFNN controller as well as the estimators of a dead-zone width and an elasto-plastic friction parameter. The validity of the proposed control scheme is confirmed from simulated results for free and constrained direct-drive robots with a deadzone in joint actuator, dynamic friction in joints and contact surfaces of the end-effector and uncertainty.