Improving Robot and Deep Reinforcement Learning via Quality Diversity and Open-Ended Algorithms - Robotics Institute Carnegie Mellon University