Optimized Containment for Multiple Unknown Nonlinear Agents via Reinforcement Learning

The optimal containment control problem is studied for a series of nonlinear multiagent systems with unknown dynamics in this work. To solve the Hamilton Jacobi-Bellman (HJB) equation associated with the unknown dynamics of agents, a reinforcement learning (RL) approch is used in design of containment algorithm with an actorcritic-identifier (ACI) architecture. The convergence is verified based on Lyapunov analysis methods. Finally, simulation studies are shown to demonstrate the control performance.