Fixed-Time Optimal Trajectory Tracking Control for an Electric Unmanned Surface Vehicle via Reinforcement Learning
Yuanbo Su, Fei Teng, Tieshan Li, C. L. Philip Chen
Abstract
Energy-saving performance and fast transient response are significant for trajectory tracking control of an unmanned surface vehicle (USV). This article proposes a fixed-time optimal tracking control scheme for an underactuated USV. First, a sensor-data-based online actor-critic (AC) reinforcement learning algorithm is proposed to derive the fixed-time optimized kinematics and actuation controllers. It can be guaranteed that the position error of the controlled USV converges to a small neighborhood of the origin in a fixed time while minimizing the cost function. Second, a class of quadratic functions composed of AC learning weights are proposed. These functions are used to demonstrate the boundedness of the estimation errors about the AC networks, facilitating the acquisition of higher order terms related to these errors. Moreover, based on Lyapunov stability criteria, the fixed-time stability of the considered USV control system is proven. Finally, the effectiveness of the proposed scheme is validated through an experiment for an electric USV considering propeller servo loop. Experimental results show that our method can achieve the optimum balance between the tracking performance and the energy consumption.