Data-Efficient Reinforcement Learning for Autonomous Robots