基于博弈模型的多智能体覆盖控制问题

Game Theory Based Coverage Control of Multi-agent

多智能体覆盖控制问题是多智能体分布式协作控制的一个重要研究方向,对于要完成给定任务的多智能体系统而言,保持整体的连通性以便信息的共享非常重要。为解决多智能体连通覆盖问题,在非合作博弈的基本模型上,提出了一种完全分布式的控制算法,即每个智能体仅利用通信范围内其他智能体的信息来独立决策自己的行动,使得系统能在基本保持全局连通的同时,实现对周围环境的最大覆盖。在决策每一步的行动时,每个智能体都希望和通信范围内的其他智能体保持尽可能大的距离,但又不能完全孤立,可看作是一个多目标组合优化问题。仿真结果表明了算法的有效性。

Multi - agent coverage control problem is an important aspect of multi - agent distributed cooperating control. For a multi - agent system having a certain task, it is important to keep it connected to share information. For solving multi - agent connected coverage control problem, the paper proposes a totally distributed control algo- rithm based on non - cooperation game theory, i.e. each multi - agent independently decides its own action only u- sing the information of other agents within its communication range, so that the system can keep connected generally as well as maximize its coverage for the environment. When an agent decides its action, it wants to be away from the others within its communication range as far as possible, however, it can be completely isolated. Obviously, this can be regarded as a multi - objective combination optimization problem. The simulation results given in the end of the paper shows the effectiveness of the algorithm.