Distributed Dynamic Clustering and Consensus in Multiagent Systems
Bosen Lian, Ahmet Taha Koru, Wenqian Xue, Frank L. Lewis, Ali Davoudi
Abstract
This paper proposes a unified framework of a distributed dynamic clustering method and a distributed control law to solve the cluster consensus problem for networked multiagent systems (MASs) with multiple leaders. Given any initial unclustered directed graph and any initial-state followers and leaders, the distributed dynamic clustering method aims to cluster each leader with at least one follower without interconnection among the clusters. This cuts some initially available edges online by considering the real-time local information of states and connections while the distributed consensus control law is running. Then, the unified framework solves the problem under mild assumptions such that all agents in the same cluster reach a consensus with the leader without interference from the other clusters. The proposed method is validated in simulation examples.