Trajectory Tracking Control for an Uncertain Mobile Manipulator: Combining Sliding Mode and Neural Network

Abstract

This paper deals with the trajectory tracking problem of a nonholonomic wheeled robot mounted with n-degree-of-freedom (n-DOF) onboard manipulator under modelling uncertainties and external load changes. Based on backstepping technique, the original problem has converted into kinematics and dynamic issues. First, the auxiliary kinematic velocity control schemes for the mobile robot and the onboard arm are developed. Second, a tracking controller, merged with the benefits of the robustness of sliding mode approach and the function approximation of neural network (NN), is presented to guarantee the velocity tracking ability in dynamic level with uncertainties. The approximated error of NN and the dynamic uncertainties are integrated into matched uncertainties. Once the system’s states are forced on the sliding surface, the robustness of the controller is obtained. All the adaptive learning algorithms of the proposed control law are derived from the Lyapunov stability theory so that the closed-loop system tracking ability can be ensured. Simulations are presented to illustrate the effectiveness of the proposed control.