Research On Cooperation Of Multi-agent System In Dynamic Environment

With the development of intelligent manufacturing,engineering control,and computer technology,multi-agent systems involving multi-agent technology have been extensively applied in many fields such as robot control and industrial production in the realm of engineering applications.Multi-agent cooperative round-up is an important research field of multi-agent system cooperation in dynamic environments,and dynamic targets round-up is a representative of multi-agent cooperative research.In a dynamic targets round-up scenario,two problems exist.One is that the existence of obstacles significantly hinders the speed and stability of the agent’s movement,and the other is how the multiagents collaboratively allocate the round-up targets when rounding up multiple targets.With the aim to dynamically round up targets,this paper focuses on cooperative task assignment in the multi-agent round-up task,intelligent obstacle avoidance in dynamic environments,and multi-target round-up methods.Firstly,a multi-objective classification round-up task assignment algorithm(MOCRTA)is proposed to solve the problem of multi-agent round-up task assignment in a dynamic environment.The algorithm calculates the distance from the agent point to each distribution point via optimal path planning result.Then the distance is used to classify the agent point,after which the classification is employed to calculate the new classification center point using distance weighting method.MOCRTA is then involved to calculate iteratively to obtain the results of the round-up task assignment.The experiment verify the effectiveness of the MOCRTA algorithm in solving the multi-agent round-up task assignment problem.Secondly,a dynamic window real-time obstacle avoidance algorithm(DWROA)is proposed for agents to avoid obstacles in a dynamic environment.The algorithm combines the path planning idea and the dynamic window method.The path planning algorithm plans the optimal static path according to the static map information,and extracts local path points.The dynamic window method performs local planning between two adjacent path points.Under the guidance of local path points,the agent quickly returns to the optimal static path after the real-time obstacle avoidance is completed,ensuring the optimal global path.For the problem of dynamic and static obstacles in the environment,DWROA improves the obstacle distance evaluation sub-function as well as the adaptability to dynamic obstacles.The experiment verify the effectiveness of the DWROA algorithm in solving the obstacle avoidance problem in a dynamic environment.Finally,a round-up potential point stage guidance algorithm(RPPSG)is proposed to solve the problem of multi-agent rounding up dynamic multi-target.The algorithm uses the MOCRTA to allocate the round-up targets based on the position of the agent,and calculates the round-up potential points for the allocation results.Then stage guidance is carried out according to the distance between the agent and the round-up potential points.DWROA is used in the long-distance state to guide the agent to avoid obstacles in real-time,while local target point guidance algorithm is employed in a close range to dynamically guide the agent.The above methods are used to guide the agent to the potential point of the roundup in stages until the roundup shape is established.On this basis,the multiagent cooperates to shrink the round-up to complete the task.The experiment verify the effectiveness of the RPPSG in solving the multi-agent dynamic and multi-target round-up problem.