Multi-ASV Coordinated Tracking With Unknown Dynamics and Input Underactuation via Model-Reference Reinforcement Learning Control
研究了利用模型参考强化学习控制方法，解决模型不确定和输入欠驱动下多艘自主水面艇的协同跟踪问题，相比基线控制和纯强化学习，算法性能更优且训练效率更高。
This article studies coordinated tracking of underactuated and uncertain autonomous surface vehicles (ASVs) via model-reference reinforcement learning control. It considered how model-reference control can be incorporated with reinforcement learning to address the challenges caused by model uncertainties and input underactuation, and how existing results may be employed to realize adaptive communication amongst ASVs. It is demonstrated that the proposed algorithm has a better performance over baseline control and effectively improves the training efficiency over reinforcement learning.