Data-Driven Suboptimal Distributed Consensus Control for Discrete-time Multi-agent Systems with Multiple Objectives

This paper develops a novel data-driven multi-agent system (MAS) consensus control algorithm. The MAS is modeled as a digraph with fixed topology. By employing the proposed algorithm, each follower agent can distributively learn a control policy online by utilizing its local information, such that the tracking consensus would be eventually achieved, and the multiple objective functions of each agent can be locally optimized in the Pareto sense. The effectiveness of the algorithm is verified by a multi-vehicle cooperative control simulation example.