摘要

The consensus problem is investigated in this paper for a class of multi-agent systems with general linear node dynamics and directed communication topologies. A new distributed observer-type consensus protocol is designed based only on the relative output measurements of neighboring agents. Compared with existing observer-type protocols, the one presented here does not require information about the relative states of the observers. Tools from small gain theory and matrix analysis, some sufficient conditions are obtained for achieving consensus in such multi-agent systems where the underlying network topology contains a directed spanning tree. Finally, some numerical examples including an application in low-Earth-orbit satellite formation flying are provided to illustrate the theoretical results.