Adaptive neural network course tracking control of USV with input quantisation and output constraints

This paper proposes a course control method for unmanned surface vehicle (USV) based on adaptive neural networks with input quantisation and output constraints. Maintaining course stability and precise control in complex maritime environments is crucial for the operation of USV. However, traditional control methods are often constrained by factors such as limited communication bandwidth, internal model uncertainties, and external disturbances. To address these challenges, firstly, we introduce a composite quantizer to describe the quantisation process linearly and then utilise a neural network system to mitigate model uncertainties and external disturbances to address these challenges, subsequently, by designing an adaptive neural network controller with input quantisation and output constraints, which not only reduces the controller's execution frequency but also ensures that control command execution remains within a safe range. By creating a Barrier Lyapunov function, the suggested control method's stability is finally shown, proving that all of the system signals eventually become confined. The system simulation results show that this methodology can enhance USV course performance, validating the effectiveness of the suggested control strategy.