Research On Multi-agent Cooperation Method Based On Intelligent Planning In Dynamic Environment

It has become an urgent need to solve distributed application in the open and dynamic environment,and the Multi-Agent System(MAS)has a unique advantage in solving this problem,so it is of great significance to research on multi-agent cooperation in dynamic environment.Most of the classic multi-agent cooperative technologies just consider the situation that environment is static,it is not suitable for solve multi-agent cooperative problem in dynamic environment.Through the dynamic cooperation between Agent,the problem solving ability of MAS can be enhanced and many problems that cannot be solved by single Agent could be done,can also adjust the behavior of MAS online according to the dynamic environment,which makes the MAS can be applied to more fields.In this paper,the cooperation of multi-agent in dynamic environment is studied,and Multi-Agent Planning(MAP)is adopted to solve the problem,the continual planning algorithm is extended to MA-PDDL,which the problem could be solved by parsing the above language.First,we apply the continual planning algorithm to MA-PDDL,and extend the syntax of MA-PDDL,which makes the language could describe the multi-agent cooperation in dynamic environment.Then,we design and implement a program that can parse the improved language,which can automatically builds multi-agent simulations and evaluate our method.Finally,a multi-agent cooperative system in dynamic environment is proposed,the MA-PDDL and the improved MA-PDDL are used to describe this system.The feasibility of the proposed method is proved by simulation experiment,which can solve the multi-agent cooperation problem in the dynamic environment.