CONSENSUS OF MULTI-AGENT SYSTEM FOR AGENTS WITH UNSTABLE LINEAR DYNAMICS

Abstract

This thesis studies the consensus problem of homogeneous multi-agent systems, where each agent is described by a general linear system. The focus is on the case where agent dynamics are unstable. The aim is to establish a set of conditions for consensus of MAS where agents have general linear unstable dynamics under various communication network settings. The adopted approach is the use of a diagonal gain matrix in a dynamic controller setting. The main contributions are: necessary and sufficient conditions for consensus for fixed networks and networks switching among undirected and connected graphs, and sufficient conditions for consensus for networks switching among directed and connected graphs and strong uniformly connected networks. These results establish the existence of the gain matrix that achieves consensus. A novel fully distributed controller based on an adaptive scheme to realize this gain matrix is also designed for the case of undirected and fixed networks.