A Trajectory Tracking Control Algorithm of Nonholonomic Wheeled Mobile Robot

Abstract

A trajectory tracking control algorithm based on deep reinforcement learning is proposed in this paper. It could solve the trajectory tracking problem of wheeled mobile robot with nonholonomic constraints. Firstly, by analyzing the nonholonomic constraint characteristics of the wheeled mobile robot, the kinematics model, dynamics model and motor drive model of the wheeled mobile robot are established. Then, according to the proposed model, a trajectory tracking control algorithm is designed by using the deep deterministic policy gradient (DDPG) algorithm. Finally, a robot agent is trained to tracking a circle trajectory by the proposed method. The simulation results show that our control algorithm could effectively track the target circular trajectory.