Unmanned Surface Vehicle Course Tracking Control Based on Neural Network and Deep Deterministic Policy Gradient Algorithm

Abstract

The paper combines the Deep Deterministic Policy Gradient (DDPG) algorithm with neural network control, which is more stable, accurate, smoother, and faster than the usual neural network control. The plan is applied to the unmanned surface vehicle(USV) course tracking control, which overcomes the shortcomings of usual intelligent algorithm learning, which requires a certain number of samples. For online learning and adjustment of the parameters of the controller can solve the uncertainty problem of ship motion in a certain extent. The simulation results show that basing on neural network and DDPG algorithm can achieve satisfactory results for the unmanned surface vehicle course tracking control, indicating that the algorithm is effective and feasible.