Multi-agent Formation Control Based On Cylindrical Topology

Multi-agent system research has always been one of the frontier issues on the international distributed artificial intelligence.Since 1990 s,with the overwhelming development of computer,control and communication technology,the scale of control system has become larger and larger,and the internal elements have become increasingly complicated.As a consequence,researchers pay highly attention to the research of multi agent system,in order to change the large-scale and complex system into a small,easy to management and research system with the function of mutual communication and coordination.Recently,the research of multi-agent system and formation coordination control mainly adopts the method of partial differential equation to modeling,which can reduce the difficulty of analysis and calculation when the system is in large-scale.Therefore,using the method of PDE to design the distributed control law of multi agent system is an effective and feasible direction.This paper proposes using PDE to build the multi-agent dynamic cluster model,and designing boundary feedback controller and observer by backstepping method,aiming at realizing the control of multi-agent system.Finally,numerical simulation is given to verify the effectiveness of the boundary controller and observer in the paper.The main research contents of this paper include the following:1.In this paper,we design the communication topology structure of the multi-agent system firstly.On account of cylindrical network topology and leader-enable formation theory,this paper sets a partial differential wave equation model on the cylindrical surface to describe the dynamic behavior of multi-agent clusters,and the state of the multi-agent system is described by the variables of the wave equation.Meanwhile,the equilibrium profile of wave equation corresponds to the formation of multi-agent system.2.Consider the instability of the equilibrium profile of wave equation,this paper needs to design a controller to stabilize the formation.It employs the method of backstepping to design the boundary feedback controller that succeeds in stabilizing the system into the desired formation.In order to realize the distributed control based on local information,we utilize the method of dual boundary controller to design the boundary observer that can estimate the state of every single multiple agents in the whole system by observing the boundary condition measurements,which are necessary for the state feedback control value.Taking advantages of the observer,the output feedback control can be achieved.3.Taking a further step to discretize the continuous partial differential equation model,we can obtain the control law of multi-agents in the system.Based on the leader-enable formation theory,the leader agents will drive the whole system into the expected formation.MATLAB simulation experiments illustrate that the method proposed in this paper can effectively control the multi-agent system from different initial positions to the desired location and realize the stable position control.