基于虚拟排斥力的移动多智能体覆盖控制动态博弈算法  

Dynamic game coverage control algorithms for multiple mobile agents through virtual repulsive force

作　　者：宿浩[1] 张宝琳 籍艳 蔡智辉 薛健 马慧[1] 郑元世 Hao SU;Bao-Lin ZHANG;Yan JI;Zhihui CAI;Jian XUE;Hui MA;Yuanshi ZHENG(College of Information Science and Engineering,Ocean University of China,Qingdao 266100,China;College of Automation and Electronic Engineering,Qingdao University of Science and Technology,Qingdao 266061,China;College of Science,China Jiliang University,Hangzhou 310018,China;School of Engineering Science,University of Chinese Academy of Sciences,Beijing 100049,China;School of Mechano-electronic Engineering,Xidian University,Xi'an 710071,China)

机构地区：[1]中国海洋大学信息科学与工程学院,青岛266100 [2]青岛科技大学自动化与电子工程学院,青岛266061 [3]中国计量大学理学院,杭州310018 [4]中国科学院大学工程科学学院,北京100049 [5]西安电子科技大学机电工程学院,西安710071

出　　处：《中国科学：信息科学》2022年第12期2195-2212,共18页Scientia Sinica(Informationis)

基　　金：国家重点研发计划(批准号:2018AAA0100804);国家自然科学基金(批准号:62273267);陕西省杰出青年基金项目(批准号:2022JC-46)资助。

摘　　要：为提高移动多智能体系统的覆盖控制效率,通常要求移动智能体在覆盖控制过程中进行多信息甚至全信息交互.在此情形下,覆盖控制成本和能耗将大幅增加.如何在提高移动智能体覆盖效率的同时,尽可能降低控制成本一直是移动智能体覆盖控制领域的重要课题之一.本文研究非完全信息条件下的移动多智能体覆盖控制方法,提出了基于虚拟排斥力的移动多智能体覆盖控制合作博弈算法.首先,将移动多智能体集群分解为若干个子集群,群中每个智能体只能观测到各自周围一定范围内其他智能体的信息,将移动多智能体集群的覆盖控制问题转化为非完全信息条件下移动多智能体子群的合作博弈分布覆盖问题;其次,给出了各移动智能体子群的博弈规则,并提出了一种基于虚拟排斥力的智能体分布覆盖博弈决策策略,用以计算并确定集群中每个智能体在覆盖控制过程中的新候选位置;第三,将移动多智能体系统覆盖控制问题分解为以刚性虚拟结构为参照系的覆盖控制问题,简化了多智能体系统覆盖控制中位移路径的规划设计.仿真算例表明,本文提出的基于虚拟排斥力的覆盖控制合作博弈算法,可以实现不完全信息条件下移动多智能体系统的分布覆盖控制任务,有效节约移动智能体的覆盖控制成本;同时,和已有的覆盖控制算法相比,本文提出的动态合作博弈覆盖算法,覆盖控制过程中智能体移动次数更少,覆盖率更高.To improve the coverage efficiency of mobile agents, the related information of the agents is generally required partially or completely during the coverage control process, which may lead to a dramatic control cost and energy consumption increase. How to balance the improvement in coverage rate and the reduction of control cost is an important issue in the coverage control area of mobile agents. This paper addresses the dynamic cooperative game coverage control algorithm design problem of a multi-agent system under an incomplete information condition. First, by separating the moving multi-agent system into several subgroups, where each agent can only obtain the position information of the neighbor agents, the coverage control problem of the moving multi-agent system is transformed into a dynamic cooperative game coverage problem of moving multi-groups under incomplete information conditions. Then, the cooperative game rules of the subgroups are presented, and a virtual repulsive force-based dynamic coverage game decision strategy of the agent system is developed to compute the new candidate positions of the agents. Third, the moving multi-agent system is considered a rigid virtual structure, and the coverage control problem of the system is turned into a new one with the virtual structure as a reference frame. Thus, the displacement path planning design of the coverage control for a multi-agent system is simplified. Simulation results reveal that the dynamic cooperative game coverage algorithm based on the virtual repulsive force can effectively realize the coverage control requirement of the moving agent system under the incomplete information condition and can remarkably reduce the cost of coverage control. In addition, compared with several existing coverage algorithms, using the dynamic cooperative game coverage algorithm proposed in this paper requires fewer moving times of agents while obtaining a higher coverage rate.

关 键 词：动态博弈 网络覆盖 分布算法 多智能体系统 虚拟排斥力 

分 类 号：O225[理学—运筹学与控制论]