Robust neural networks trajectory tracking control with zero-sum game thought for unmanned surface vehicle under marine ranching cruise task

To promote the marine ranching automation and ocean cybernetics, this paper focuses on multiple net cage encirclement and robust control problems of unmanned surface vehicle (USV). In the proposed strategy, a guidance law is developed for leading the USV to perform marine ranching tasks autonomously, where a waypoints path connects all the net cages. The tasks of net cage encirclement and waypoints path navigation can be switched by using the proposed guidance law. Next, a control algorithm is designed for USV to track the guidance signals. Specially, a zero-sum game based anti-disturbance approach is proposed to achieve better tracking performance and robustness. Besides, the nonlinear uncertainties and unknown actuator gains are also compensated by introducing neural networks and adaptive laws. The designed controllers are “user-friendly” and conform to nautical engineering practice. The stability of the proposed closed-loop control system is proved satisfying semi-globally uniformly ultimately bounded stability. Finally, two simulation cases are illustrated to verify the superiorities of the proposed algorithm.