Research And Implementation Of Multi-Agent System Formation Collision Avoidance Algorithm Considering Physical Size

In recent years,with the development of large-scale embedded computers,the application scope of agent systems in industry and military has gradually expanded.People gradually realize that through coordination and cooperation between agents,they can complete more complicated products of a lower cost.The task,because the multi-agent system contains distributed sensing and execution systems,and its inherent parallelism,has strong redundancy,and has better robustness and fault tolerance than a single agent.Therefore,the multi-agent formation system has developed into one of the important research directions in the field of group intelligence and control.This paper analyzes the multi-agent system formation algorithm proposed to home and abroad,discusses the superiority in centralized,distributed and decentralized formation algorithms,and combines the virtual structure method,artificial potential field method to propose two multi-agent formation collision avoidance algorithm.The challenging problem is the design of the intelligent body system considering the physical size and the real-time painting to realize the dynamic evolution process of the system.The research on the subject has achieved the results as follows.1.The circular formation and triangle formation algorithm of the second-order integra tor multi-agent system is designed by the virtual structure method,and its feasibility and e ffectiveness are discussed.2.The physical size design of each agent in the agent system collision avoidance algorithm is completed.The collision avoidance part algorithm design of multi-agent system formation is considered considering the actual size of the individual.The artificial potential field functions are used to construct the collision avoidance detection boundary and the collision avoidance area.It limits the perception ability of the agent,realizes the algorithm setting from centralized to distributed,reduces the degree of coupling with individuals in the system,and improves the robustness of the formation system.The individual collision avoidance algorithm considering physical size effectively prevents the individual from colliding behavior due to collision.3.A multi-agent system formation simulation framework based on Processing is designed.Based on the eclipse application Java language,the secondary development of the simulation platform enhances the readability of the algorithm code,and the application design language object-oriented,inheritance and other ways to achieve multi-intelligence.The module design of the body formation algorithm and the MATLAB implementation of the multi-agent system formation algorithm considering the physical size of the three-dimensional mode is also proposed.4.The design of the multi-agent dynamic model is completed.In order to improve to the traditional trajectory display scheme for intelligent body,referring to the bionic robotic fish swimming model of the intelligent control laboratory platform of Peking University,An intelligent bionic fish model based on the third-order Bezier curve for soft swimming is proposed,combined with bionics.Simulation and analysis of the fish body motion model and multi-agent formation collision avoidance algorithm verify the feasibility and effectiveness of the algorithm.Finally,using the new neural network reinforcement learning model,combined with the unicycle formation plan,using the Follower-Leader formation mode,a formation collision avoidance scheme is proposed,and the DQN Network mode is applied.In the case of multiple training,the intelligence can be made.In the process of formation,the body effectively avoids the collision of obstacles to the environment,and finally gives the trajectory of the unicycle and the convergence from the neural training.