not a paper that came out this year but Emergence of Locomotion Behaviors in Rich Environments by Google deepmind: (https://arxiv.org/abs/1707.02286 https://arx

not a paper that came out this year but Emergence of Locomotion Behaviors in Rich Environments by Google deepmind: (https://arxiv.org/abs/1707.02286 https://arxiv.org/abs/1707.02286) made me realize that although RL is brittle and slow it can still be super effective the problem domains I care about (robot locomotion).