摘要
随着无人艇自主控制技术的发展,其在军事领域的作用日益突出,反水雷无人艇的自主控制技术是目前研究的热点之一。针对反水雷无人艇的局部路径规划问题进行研究,提出一种分层强化学习方法,对作为无人艇路径规划器的进化神经网络进行训练。同时使用Unity物理引擎搭建仿真环境,构建了具有环境感知和自主决策能力的无人艇模型。试验验证表明了算法在处理局部路径规划问题上的有效性。
Unmanned surface vehicles are finding more and more prominent applications in military field with the development of autonomous control technology.The intelligent control technology of Mine Countermeasures Unmanned Surface Vehicle(MCM USV) is one of the current research hotspots.Aimed at the local path planning problem of MCM USV,a hierarchical reinforcement learning method is proposed,which is used for training the evolutionary artificial neural network used as path planner of the UAV.The Unity physics engine is adopted to build a simulation environment,and a USV model with environment awareness and autonomous decision-making capabilities is established.Experimental result verifies the effectiveness of the algorithm on local path planning.
作者
杨全顺
尹洋
陈帅
YANG Quanshun;YIN Yang;CHEN Shuai(Naval University of Engineering,Wuhan 430000,China)
出处
《电光与控制》
CSCD
北大核心
2021年第7期11-15,共5页
Electronics Optics & Control
基金
国防“九七三”资助项目。
