Fully distributed data-driven model-free adaptive control for consensus tracking in multi-agent systems

Abstract

This paper introduces a fully distributed model-free adaptive control (MFAC) approach for consensus tracking in multi-agent systems (MASs) with compact form data linearization (CFDL). Unlike prior methods that require agents to know the full communication graph, our approach allows each agent to configure its controller using only local information from its neighbors, achieving a fully distributed control. Therefore, our method easily supports scenarios where agents dynamically join or leave MAS. Additionally, our approach does not require a strongly connected communication graph and consensus can be achieved as long as the graph includes a spanning tree with the leader as the root. Simulations demonstrate that this method converges faster to the desired trajectory compared to previous MFAC-based methods.