Haiying Wan

This paper investigates a hierarchical residual reinforcement learning framework for nonlinear systems with piecewise disturbances. To address partial observability problems caused by piecewise disturbance switching, a piecewise disturbance inference mechanism is introduced. Based on this inference mechanism, a hierarchical policy architecture is developed, decomposing the control task into a base policy and a residual policy.

Abstract Uncrewed underwater vehicles(UUVs) play an indispensable role in ocean resource efficiency utilization due to their security and cost-effectiveness. However, the significant uncertainties, time-varying disturbances, and unstructured subsea environments pose challenges for UUVs in achieving precise and efficient marine missions.

Conflicts of Interest The authors declare no conflicts of interest. References 1, , , , and , “Data-Driven Event-Triggered Adaptive Dynamic Programming Control for Nonlinear Systems With Input Saturation,” IEEE Transactions on Cybernetics 54, no. 2 (2024): 1178–1188. 2, , , , and , “Mismatched Quantized H∞ Output-Feedback Control of Fuzzy Markov Jump Systems With a Dynamic Guaranteed Cost Triggering Scheme,” IEEE Transactions on Fuzzy Systems 32, no. 4 (2024): 1681–1692.