Learning-Based Bipartite Output Consensus for Asynchronously Switched Multi-Agent Systems with UAV Payload Transport Applications

This paper focuses on achieving output consensus for switched multi-agent systems with mixed cooperative and antagonistic interactions under completely unknown leader dynamics. Firstly, the considered model is more general, accommodating asynchronous switching signals, completely unknown leader system matrices, and the simultaneous coexistence of cooperative and antagonistic interactions, thus relaxing critical limitations of prior works. Secondly, using a data-driven approach, this paper develops a learning framework to estimate the unknown leader system matrix from operational data. Thirdly, this paper proposes a novel switching control strategy that ensures bipartite output consensus by designing agent-dependent switching laws and constructing switching-dependent controllers incorporating independent virtual reference generators, based on agent-dependent average dwell time conditions. Finally, the effectiveness of the theoretical approach is validated through payload transportation scenarios employing unmanned aerial vehicles (UAVs).