一种动态环境下多Agent的协作方法

摘要

对动态环境下多Agent系统中Agent的协作问题进行了研究,提出了一种基于连续规划的动态环境下多Agent协作方法.首先,在MA-PDDL的语法上进行了改进,通过一个自定义的函数使其能够描述连续规划方法;然后,提出了一种基于扩展的MA-PDDL的连续规划算法,能够处理动态环境在系统中所造成的不确定性;最后,实现了扩展的MA-PDDL的解析方法,解析的结果能够用于模拟Agent执行任务的过程.选取了医疗垃圾无人运输场景进行实验,成功模拟了整个实验场景的运行过程,验证了方法的可行性.%A multi-agent cooperation method was proposed based on multi-agent system continual planning in dynamic environment.First,the Multi-Agent Planning Domain Definition Language (MA-PDDL) was extended for describing the continual planning method by a user-defined function.Then,a continual planning algorithm based on extended MA-PDDL was proposed to deal with the uncertainty.Finally,a parsing method of the extended MA-PDDL was implemented,which simulated how agents execute tasks.The method successfully simulates the execution process of automated cart transportation in hospital,which validates its feasibility.