Improved Line-of-Sight Path Following Control for Underactuated USVs with Unknown Parameters Using Q-learning

This paper proposed an intelligent path-following control strategy that combines the line-of-sight (LOS) guidance law and Q-learning (QL) for an underactuated unmanned surface vehicle (USV) with unknown model parameters. The improved LOS achieves satisfactory path-following results for USVs with uncertain models, where the embedded QL method learns the vehicle model by the self-regulation of uncertain parameters to accomplish the robustness and adaptation for the path-following. Simulation results in tracking a straight line and a circular path are demonstrated respectively. The environmental effect of wave disturbances is also analyzed, which verifies the effectiveness of the improved LOS method based on QL.