徐勇研究员，北京理工大学

This talk focuses on the even-triggered cooperative control problem of heterogeneous multi-agent systems (MASs) using data-based reinforcement learning (RL) algorithm. To lower the communication and computing burden among agents, an event-driven adaptive distributed observer and corresponding integral input-based triggering mechanism are proposed. To obtain the optimal controller while avoiding the requirement of relying on systems dynamics, a RL-based controller using model-based control policy is developed, which is further extended to model-free case without requiring any knowledge of the system dynamics. Rigorous analysis shows that the developed algorithms not only save the limited network resources, but also obtain the output tracking in an optimal fashion. Finally, a numerical simulation is demonstrated to verify the proposed approach in theory.