An adaptive path-tracking framework with velocity adaptation for underactuated AUV in large curvature paths

Accurate large-curvature path tracking is significant for underactuated autonomous underwater vehicles (AUVs) to perform tasks effectively in confined environments. To enhance the tracking performance under large-curvature paths and time-varying disturbances, this study proposes an adaptive large-curvature path tracking framework based on surge velocity adaptation and broad learning system-based adaptive controller. A curvature-adaptive line-of-sight (LOS) is firstly developed through an adaptive virtual target updating mechanism and look-ahead distance. Then, an optimisation-based surge velocity regulator is designed to dynamically regulate the surge velocity in response to path curvature variations. Leveraging the excellent nonlinear fitting capabilities of broad learning system (BLS), it is introduced into the sliding mode controller to approximate unknown dynamics induced by external disturbances. The stability analysis based on Lyapunov method proves that the tracking errors and estimation error of the closed-loop system are uniformly ultimately bounded (UUB). Comparative simulations show that the maximum tracking error of the proposed method is about 34% for simple paths and 50% for large-curvature paths lower than the compared algorithms, and the ablation study validates the indispensability of the components of the proposed method. These results validate the effectiveness and superiority of the proposed framework in enabling precise large-curvature path tracking of underactuated AUVs.