Robust Bipartite Output Containment of Heterogeneous Non-introspective MAS on Signed Digraphs

Abstract

Output containment is achieved among linear heterogeneous multi-agent systems (MAS) thanks to the collaborative efforts of all agents, expressed by a communication digraph with nonnegative weights. Generally, state-feedback or output-feedback designs are used in the distributed containment control of MAS, which require the absolute value of state/output of each agent. This paper investigates the output containment of heterogeneous non-introspective MAS on signed digraphs, in the presence of uncertain parameter variations. Nonintrospective agent only has knowledge of relative information with respect to its neighbors, rather than the explicit knowledge about its own state/output. Agents on signed digraphs can have antagonistic interactions, modeled as negative weights on the communication network. To this end, we first formulate a new control problem called the robust bipartite output containment problem (RBOC), which aims at making each follower's output converge to the dynamic convex hull spanned by the outputs and the negative outputs of the leaders. It is proved that the RBOC problem can be solved by making certain signed output containment errors go to zero asymptotically. Then, a dynamic output-feedback control protocol is designed based on internal model principles. Finally, local sufficient conditions are obtained and explicit local design procedures are provided. Numerical simulations are performed to illustrate the proposed RBOC performances on signed communication digraphs.