Research On Differential Evolution For Multi-agent Path Planning

With the development of industrial technology,the problem of agent path planning needs to be completed in a more complex and diverse working environment.However,due to the limited application scope of single agent,multi-agent system is needed to improve its performance.The path planning of multi-agent is a complex problem,and it is necessary to consider the technical theory of environmental modeling,obstacle avoidance and coordination.In this paper,we improved the Differential Evolution(DE),and studied the path planning problem of single / multi-agent under the known condition of global information.The main work of this article includes:(1)The research situation of the intelligent agent path planning and the related research status at home and abroad are introduced.The common methods of path planning are introduced,and the advantages and disadvantages of these methods are analyzed.The research status of DE at home and abroad,the commonly used improvement strategies,and the basic principles and processes of the algorithm are introduced.(2)Aiming at the advantages and disadvantages of the existing environmental modeling methods,we establish the environmental model of this paper.Under this environment model,we design the path encoding and the shortest collision free path model.(3)A hybrid algorithm by combining artificial potential field with differential evolution is proposed for solving the collision free shortest path planning problem of mobile robot.First,collision free path model is built for the global environment information of mobile robot.Meanwhile the optimal path of robot is planned by differential evolution algorithm.Adaptive adjustment strategy is used for the ariation factor of differential evolution.For the infeasible solution appeared in the crossover operation of differential evolution,an artificial potential field method is proposed to improve the efficiency of the algorithm.The experimental results show that,the presented method can plan a shorter and more reasonable path than traditional differential evolution algorithms,which represents better convergence speed and the solution quality,and effectively solve the collision free path problem.(4)The MDEMEA algorithm is proposed and applied to the multi-agent path planning problem.The initial population of DE is divided into multiple subpopulations of the same size.According to the corresponding mathematical model,the MDEMEA algorithm is used to plan the multi-agent path,and the Multi-Agent Coordination is designed for the collision path points between the agents produced in the algorithm.The simulation results show that the proposed MDEMEA algorithm is practical.