Fetching the paper…

Keypoints into the Future: Self-Supervised Correspondence in Model-Based Reinforcement Learning · Around