Deepak Pathak Demonstrates In-Context Learning for Robotics with One-Shot Video Prompts

deepakpathak · x · 2026-08-26

Deepak Pathak demonstrated in-context learning capabilities for robotics. The method allows robots to perform long-horizon tasks exceeding 10 minutes using just a single video prompt, without fine-tuning and without seeing the tasks during pre-training. This approach aims to build intelligence from the foundations up.

Original post →

More from Embodied

Embodied channel →