Zero-shot performance

This research introduces Robotic Policy Inference via Synthetic Observations (ROSO), a novel method to improve the performance of pre-trained policies in robotic systems. ROSO uses a technique called Stable Diffusion…