[BREAKTHROUGH2024] How to Train Your Robot: Large Behavior Models and Visuomotor Diffusion Policy