Optimal Path Seeking for Consensus Tracking Control of Networked Marine Surface Vehicles: A Three-Layer Reinforcement Learning Approach

This study introduces a novel three-tiered reinforcement learning (RL) framework, meticulously crafted for the purpose of optimal pathfinding and consensus tracking control within the realm of networked marine surface vehicles (NMSVs). The framework is structured into three layers: decision-making, estimation, and control, each addressing specific challenges in autonomous marine navigation. The decision-making layer processes environmental data and employs reinforcement learning for path planning, the estimation layer develops consensus strategy development, and the control Layer implements these strategies for coordinated MSV movement. Our experimental results, obtained from simulations, demonstrate effectiveness for the framework in improving operational efficiency and reliability of MSVs in dynamic marine environments, underscoring the potential of our approach in advancing autonomous marine vehicle coordination.