Event-Driven Intelligent Dynamic Positioning for Networked Unmanned Marine Vehicles Within Reinforcement Learning Framework
Jian Liu, Jiachen Ke, Jinliang Liu, Xiangpeng Xie, Engang Tian · IEEE Transactions on Vehicular Technology · 2024
This correspondence develops an intelligent dynamic positioning regulation strategy for nonlinear networked unmanned marine vehicle (NUMV) with unmatched interference. Distinct from the existing Takagi-Sugeno fuzzy approach, reinforcement learning (RL) framework with simplified parameters is utilized to pursue the regulation optimality. On basis of adjustable weight and threshold function, we provide a novel adaptive weight event-driven scheme (AWEDS) to achieve the efficient information transmission. Furthermore, an AWEDS-boosted critic-sole iterative RL algorithm is proposed for implementing the approximately optimal control policy. Afterwards, the nonlinear NUMV and the critic weight estimation error are ensured to be uniformly ultimately bounded. The feasibility and validity of the proposed learning algorithm are eventually illustrated by a comparative experiment.