Improving Robot And Deep Reinforcement Learning Via Quality Diversity, Open-Ended, And...