1 paper
Maxime Alvarez, Ryo Watanabe, Paul Crook +4
As end-to-end robotic policies are progressively deployed in the real world to solve real tasks, they face a gap between the training and inference conditions. Scaling the amount a…