Motion planning of a quadrotor robot game using a simulation-based projected policy iteration method