Abstract: We consider an off-policy reinforcement learning algorithm that gathers input and state data from a nonlinear system, and uses them to approximate the infinite-horizon optimal control for ...