Teaching-a-computer-to-strafe-jump-in-Quake-with-reinforcement-learning