Resilient Navigation Control for Autonomous Ships Under Complex Environmental Disturbances

To enhance the resilience of Maritime Autonomous Surface Ships (MASS) under complex environmental disturbances, this paper proposes a hierarchical decision-making framework that decouples global path planning and local obstacle avoidance. The global policy utilizes structured navigational states for long-range trajectory generation, while the local policy leverages radar-based perception for fine-grained control. A high-fidelity simulation environment is developed with wind–wave–current disturbances and dynamic obstacles, alongside a physics-informed composite reward function to ensure control stability and interpretability. Experiments demonstrate that the framework achieves an average success rate of 91.2% in disturbance-rich conditions and maintains over 85% in high-risk scenarios. Path efficiency exceeds 0.65, and rudder smoothness remains below 0.2. The results validate the proposed method’s ability to sustain navigation continuity and control robustness, confirming its potential for resilient MASS deployment in dynamic maritime environments.