Gaussian Process-Based Integral Sliding Mode Control for Robust AUV Trajectory Tracking

This paper proposes a novel robust control method for autonomous underwater vehicle (AUV) trajectory tracking, specifically addressing challenges arising from sudden changes in desired trajectories and significant modeling uncertainties. The method leverages sparse online Gaussian process (SOGP) to learn unknown dynamics online, given a prior AUV model. An exponential forgetting scheme is integrated into the SOGP algorithm to ensure the selected training dataset incorporates both temporally and spatially informative data. By incorporating Gaussian process (GP) posterior information into an integral sliding mode control (ISMC) framework, the proposed GP-ISMC scheme enhances robustness and adaptability to uncertainties and disturbances encountered during diverse motion patterns. Lyapunov analysis guarantees system stability and tracking error convergence with a specific probability. Simulation results demonstrate that the proposed method has superior performance in tracking accuracy and prediction reliability.