A Machine Learning-Based Parameter Adaptive S-Plane Controller for Motion Control of Unmanned Underwater Vehicles

The complex dynamic characteristics and marine environment pose significant challenges to the motion controller design of unmanned underwater vehicles (UUV). However, the controller parameters are usually fixed once manually adjusted, so they cannot adapt to environmental changes. This paper aims to present a machine learning-based parameter adaptive S-plane controller for the motion control of UUVs. The Q-learning-based adaptive control method is adopted to realize the optimization and automatic tuning of controller parameters in different environments. The optimal mapping between the input state and output action is found through the self-learning mechanism of Q-learning. Simulation results indicate that the proposed controller can adjust the controller's parameters in real-time and online and has a good control effect and adaptive ability for the actual environment.