Leader Selection And Controllable Containment Control Of Multi-agent Systems

In recent years, coordination control of multi-agent systems has become a central issue in control field with the rapid development of modern science and biology、computers、artificial intelligence and other cross-disciplinary in-depth exploration, it also successfully used in multi-robot cooperation, traffic flow control, unmanned aerial vehicles formations and other machine engineering. According to the different number of leaders in multi-agent systems,consensus problems can be divided into the following situations: leaderless consensus,consensus tracking(leader-following consensus), multiple leader consensus. As a special multiple leader cooperative control, the containment control has become one of the important research contents of the multi-agent cooperative control filed.This paper research the controllable containment control of multi-agent systems based on studying and summarizing the related research results and the relevant principles of control theory. The main research contents and contributions are as follows:1. The controllable containment control for second-order under given topology directed multi-agent network is investigated in this paper, Aiming at most of the results of containment control, the controllable configuration issues have not been considered between leaders and followers with multi-agent networks, The complex network controllability theory and the maximum bipartite graph matching algorithm are combined to determine the leader and follower set meeting the network controllable. Using the designed corresponding control protocol for followers, all followers are driven to converge to the stationary convex hull with the leaders in order to achieve the controllability containment control of the networks.2. The leader selection and controllable containment for given topology that has strongly connected sub-graph of directed multi-agent systems is investigated. According to the topology structure, the agents are classified into two categories: unit agents and general agents.Firstly, the consensus protocol of unit agents in the strongly connected graph is designed to achieve consensus of each unit; Secondly, combining with the maximum bipartite graph matching algorithm, the new topology constituted by units and general agents to determine the minimal leader set meeting the network controllable. By using the designed corresponding control protocol for all agents, all followers are driven to converge to the convex hull with the leaders in order to achieve the controllable containment control of the networks.3. Considering the existing containment control algorithms in multi-agent system cannot be efficient for task distribution when implement multi-objective task and satisfied the system controllable. In this paper, we first optimize the global topology of the multi-agent system combined with community partition algorithm, and then the maximum bipartite graphmatching algorithm is combined to determine the leader set meeting the community controllable. Meanwhile the distributed tasks are given to the leader of each community within the tasks completed controllable. After that according to the topology structure between communities, applying the previous methods to find the leaders communities satisfied the network controllable, by using the designed estimate control protocol for all communities to achieve the controllable containment control.