Research On Consensus-based Formation And Collision Avoidance Of Multi-agent System

As a distributed system with multiple agents interacting with each other,multi-agent system has been more often studied by many researchers and institutions in recent years.Compared to a single-agent system,a multi-agent system has a clear advantage in performing tasks in complex and dangerous environments,with better flexibility,lower cost,and greater survivability.As an important part of the autonomous collaborative control of multi-agent system,the formation controller make the system converge to the pre-designed formation by constraining the control input,which facilitates the information interaction between the agents and improves the capability of environmental situational awareness of the system.Meanwhile,in the process of formation or maintenance of formation,it is also necessary to avoid collisions that may occur between agents caused by too close displace.In this paper,problems of formation of single-order linear and second-order nonlinear multi-agent system are studied,and problem of collision avoidance of single-order linear multi-agent system is studied.The main content and innovations in this paper include:(1)A formation control algorithm of single-order linear multi-agent system with fixed topology is designed based on consensus theory.And the global asymptotic stability is proved based on Lyapunov theory.Simulation experiments on Matlab have verified the effectiveness of the controller.(2)For the second-order nonlinear multi-agent system with a switching topology,a new consensus-based formation controller is designed and the global convergence proof is provided.The simulation experiment shows that the controller performs well.(3)For the first-order linear multi-agent system,a formation controller with collision avoidance capacity is designed based on consensus theory and artificial potential field theory.An adjustment parameter ? is introduced to ensure that the connectivity between the agents is not lost due to the excessive repulsion of the potential field.At the same time,the application of model predictive control in trajectory tracking is studied.On this basis,the collision avoidance algorithm procedure is designed,and the idea of collision avoidance of multi-agent is expanded.