多种会遇状态下基于强化学习的船舶自动避碰路径仿真  

Ship Automatic Anti-collision Path Simulations Based on Reinforcement Learning in Different Encounter Situations

作　　者：赵舟 王俊雄[1] ZHAO Zhou;WANG Jun-xiong(School of Naval Architecture,Ocean and Civil Engineering,Shanghai Jiaotong University,Shanghai 200240,China)

机构地区：[1]上海交通大学船舶海洋与建筑工程学院,上海200240

出　　处：《科学技术与工程》2018年第18期218-223,共6页Science Technology and Engineering

摘　　要：为实现海域中多会遇局面的船舶自动避碰,依据船舶间状态信息的实时获取,构建了自适应启发评价的船舶避碰算法。算法中通过再励信号由强化学习训练得到船舶从会遇形成到驶过让清动态过程的优化转向决策,模拟给出了两船会遇局面、多船会遇局面下的避碰优化路径。结果表明算法能够满足避碰规则并完成安全避碰,与已有分布式避碰决策方法对比,显示了更好的实时性和经济性。On the basis of real-time state information between ships,an adaptive heuristic critic algorithm for ship automatic anti-collision was presented in order to deal with different encounter situations.In the algorithm,the optimization decision was formed by reinforcement learning trained in the whole dynamic process.The simulations of ship collision avoidance paths were presented in different encounter situations.Results show that give-way ships are complied with COLREGs and implement secure anti-collision.Compared with the existing distributed anti-collision decision,it shows better real-time operation and economy.

关 键 词：船舶自动避碰 强化学习 自适应启发评价 避碰路径 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]