Neural network-based tracking control of autonomous marine vehicles with unknown actuator dead-zone

This article studies the neural-network based backstepping control problem for autonomous marine vehicles perturbed by external disturbances. The actuator dead-zone phenomenon, which is a common non-smooth property caused by the complicated operating environment of autonomous marine vehicles, is also considered. To cope with the issue of "complexity explosion" and further decrease the tracking errors, a command filtering compensation strategy is also proposed, which guarantees satisfactory tracking performance and boundedness of the closed-loop system signals. Finally, simulation studies are given to further demonstrate the effectiveness of the proposed control method.