Optimal Model-Free Output Synchronization of Heterogeneous Multiagent Systems Under Switching Topologies

In this article, the output synchronization of discrete-time heterogeneous multiagent systems over directed switching topologies is investigated. For output synchronization problem, the optimal control protocols for all agents depend on the solutions of a series of algebraic Riccati equations (AREs), which are difficult to be solved analytically. Besides, both the agents' and the leader's dynamics are supposed to be unknown. The distributed adaptive observer for each agent is designed first to estimate the leader's state without requiring complete knowledge of the leader's dynamics. Based on the trained observers, the output synchronization problem is formulated to an optimal control problem. An observer-based Q-learning algorithm is developed to solve the AREs using system data rather than the accurate system models. The optimal analytic distributed control policies can be obtained by policy iteration combined least square method. It is proved theoretically that the output synchronization is ensured based on the distributed adaptive observers under switching topologies. Ultimately, the theoretical results are demonstrated via three simulation examples.