New Directions in Reinforcement Learning and Control