复杂约束条件下异构多智能体联盟任务分配  

Coalition task allocation of heterogeneous multiple agents with complex constraints

作　　者：唐嘉钰 李相民 代进进 薄宁 TANG Jia-yu;LI Xiang-min;DAI Jin-jin;BO Ning(College of Coastal Defense,Naval Aviation University,Yantai Shandong 264001,China;Unit 91213 of the PLA,Yantai Shandong 264001,China)

机构地区：[1]海军航空大学岸防兵学院,山东烟台264001 [2]中国人民解放军91213部队,山东烟台264001

出　　处：《控制理论与应用》2020年第11期2413-2422,共10页Control Theory & Applications

基　　金：国家自然科学基金项目(61305136)资助.

摘　　要：分布式任务决策是提高多智能体系统自主性的关键.以异构多智能体协同执行复杂任务为背景,首先建立了一种考虑任务载荷资源约束、任务耦合关系约束及执行窗口约束等条件的异构多智能体分布式联盟任务分配模型;其次,对一致性包算法(CBBA)进行了扩展,提出了基于改进冲突消解原则的一致性联盟算法(CBCA),以实现异构多智能体协同无冲突任务分配,并进一步证明了在一定条件下CBCA算法收敛于改进顺序贪婪算法(ISGA).最后通过数值仿真,验证了CBCA算法求解复杂约束条件下异构多智能体联盟任务分配问题的可行性和快速性.Distributed decision making is the key of enhancing the autonomy of multi-agent systems.Firstly,based on the problem of multi agents cooperating to perform complicated tasks,a decentralized coalition task assignment model of heterogeneous multiple agents with consideration of specific equipment requirements of tasks,coupled constraints and the time window constraint is proposed.Secondly,the consensus-based bundle algorithm(CBBA)is extended to the consensusbased coalition algorithm(CBCA)with the improved principles of conflicts resolution to achieve the conflict-free solution of the task assignment problem.Further CBCA is proved to be convergent to the improved sequential greedy algorithm(ISGA).Finally,simulation results of multiple unmanned search and combat aerial vehicles performing search ad attack missions demonstrate the feasibility and rapidity of the proposed algorithm.

关 键 词：多智能体系统 分布式决策 一致性包算法 联盟生成 无人机 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]