Keynote 2 Improving Robot And Deep Reinforcement Learning Via Quality Diversity, Open Ended, And A