Robust adaptive optimal control for unknown dynamical systems

In this paper, we propose an algorithm of adaptive optimal control scheme for systems whose dynamics are unknown and the states are contaminated by noises. The basic control law is Policy Iteration which can solve HJB equation recursively. In the proposed method, the value function is estimated using a nonlinear filtering but the state of the system is not estimated since the system model is not available. Since the proposed method can reduce the effects of the noises without using the system model, we can apply this method to many practical systems without model and parameters.