The Coordination Control Methods And Experimental Research On The Multi-agent System To Perform Tasks

As a new artificial intelligence technology,multi-agent system is more and more popular.In order to perform various types of tasks,integrating system features and improving key technologies is of paramount importance.Therefore,this paper solves three key problems: task assignment,path planning and formation control.Firstly,to solve the task allocation problem with timing characteristics,the mathematical model is established and the continuous particle is discretize to solve the optimization problem.Particle position is divided into two parts,where the integral part represents the task and the remainder part represents the priority of the task.Experiments show that this method can balance the number of tasks and reduce the execution time.Secondly,in order to find the shortest path,the environment model is established by using the grid method,and ant colony algorithm is improved by non uniform initial distribution of pheromone,increasing with direction selection strategy,covering update pheromone update,segmented pheromone evaporation coefficient adjustment.The experimental results show that the improved algorithm can better balance the convergence and rapidity,and theoretically prove that the probability of searching the optimal solution tends to 1.Finally,the leader follower method is adopted to solve the formation problem.Based on multi-agent discrete time model,introducing the real-time movement speed of adjacent agents can complete formation control based on the original controller to complete formation.According to Routh criterion parameter range is determined.Experiments show that the improved controller can realize the formation task faster,and the noise effect is far less than the original controller.