Optimal Bipartite Consensus Control for Nonlinear Multiagent Systems Based on Reinforcement Learning

This paper investigates the optimal bipartite consensus control problem for nonlinear multi-agent systems (MASs) with unknown dynamics. First, a policy iteration (PI) method using the formulated Q-function is proposed and rigorously analyzed to demonstrate its effectiveness in obtaining the optimal control policy. Second, the algorithm employs Integral reinforcement learning (IRL) to implement the model-free optimal bipartite consensus control, and an experience-based weights update law is designed to enable the tuning law to reuse history data in weights updates. Finally, the effectiveness of the proposed method are verified by simulations.