Cooperative Differential Graphical Game Theoretic for Tracking Control of Nonlinear Multi-Agent Systems With Unknown Dynamics

This paper investigates the cooperative tracking control problem of networked nonlinear multi-agent systems (MASs) with completely unknown dynamics. By formulating the optimal tracking control problem into a cooperative differential graphical game, we can employ the off-policy integral reinforcement learning (IRL) scheme to find optimal tracking controllers online along with the system trajectories without requiring the knowledge of the system dynamics. In contrast to the existing literature where the Nash equilibrium is utilized to characterize the performance of the designed controllers for the cooperative control of MASs, we introduce a new solution concept regarded as Pareto optimality strategies which devote to minimize performance cost and risk of all agents simultaneously. A simulation example is presented to verify the effectiveness of the proposed approach.