Research On Multi-Agent Cooperative Collision Avoidance Method

An agent is an entity that has the ability of independent thinking and can obtain environmental information.An agent has a certain ability to perceive the surrounding environment and can make corresponding changes according to the changes of the environment.Agents can coordinate and interact with each other,follow the rules to move toward the goal through the given goal,and guide other agents to move forward.Multi-agent cooperative obstacle avoidance is the study of how a group of agents cooperate to form a stable system and avoid obstacles to reach a specified location.With the rapid development of modern robot technology,multi-agent cooperative obstacle avoidance is particularly important in the field of robot research,and has been widely used in sports competition,military,medical,agriculture,industry and other fields.Although there have been a large number of cooperative collision avoidance algorithms and achieved certain results,these algorithms still have their own shortcomings.In terms of formation,each formation algorithm has its unique characteristics and formation mode.In terms of obstacle avoidance,different obstacle avoidance algorithms need to be considered according to different scenarios.Therefore,this paper combines the leader-following method,graph theory and consensus theory to study the local path planning and global path planning in the multi-agent case,respectively.Firstly,aiming at the formation and maintenance problems of multi-agent formation,this paper proposes to combine the navigation and following method with graph theory,and uses the consistency theory to complete the formation control algorithm of multi-agent formation.Meanwhile,the pentagram formation and regular hexagon formation are designed to verify the effectiveness and expansibility of the formation algorithm.Simulation experiments show that after a certain number of iterations of the algorithm,the multi-agents can complete the desired formation and keep the formation moving towards the target point,and finally reach the target point.Secondly,aiming at the problem of local path planning in the case of multi-agents,artificial potential field method is proposed to complete obstacle avoidance on the basis of formation algorithm,the shortcomings of traditional artificial potential field method are studied and improved,and the target point distance and random velocity value are introduced at appropriate time to solve the problem of unreachable target and local minimum of the original algorithm,simulation results show that the improved algorithm is effective.Finally,aiming at the problem of global path planning in the case of multi-agent,a random intelligent search algorithm named RRT(Rapidly-expanding Random Tree)algorithm was studied.In order to solve the problems of too random RRT algorithm,rough planning path and low overall efficiency of the algorithm,an improved bidirectional RRT algorithm based on goal guidance was proposed.Aiming at the problem of formation transformation when the multi-agent encountered obstacles,three modes including zero transformation mode,homogeneous transformation mode and heterogeneous transformation mode were introduced to make the multi-agent pass the obstacle area while maintaining the original formation as much as possible.Finally,the simulation results show that the improved algorithm performs better than the original algorithm.