An online adaptive optimal control scheme based on identification-evaluation architecture is presented for the optimal coordination control of uncertain, non-linear, multi-agent systems.In light of systematic uncertainty, an identification Neural Network (NN) is utilized to each agent to approximate the uncertain system dynamics, and an evaluation NN is used to estimate the solution of the coupled Hamilton-Jacobi (HJ) equations.Then, the optimal control law can be derived.Based on the developed architecture, the weights of the identification NN and the evaluation NN can be updated synchronously. By using Lyapunov's direct method, it is guaranteed that the weight errors of the identification NN and the evaluation NN are uniformly and ultimately bounded, and that the closed-loop system is stable.Finally, a numerical example is provided to demonstrate the effectiveness of the proposed control scheme.