Google DeepMind: Gemini 4 Argon Powers Humanoid Video-to-Sim-to-Real Pipeline
In an internal test, Google DeepMind robotics researchers instructed Gemini 4 Argon to analyze a video of a human playing cornhole. Utilizing its video understanding and advanced reasoning, Gemini 4 constructed a virtual simulation inside MuJoCo. The model then calculated the physics and whole-body control parameters necessary for physical execution. Operating with these calculations, Apptronik's Apollo 2 humanoid robot sank seven consecutive cornhole shots in the real world. Google DeepMind highlights that this Video-to-Sim-to-Real pipeline compresses weeks of manual modeling into hours, turning ordinary video into a training resource for physical AI.
Original sources
YouTube Google DeepMind
Video:Teaching a humanoid robot to play cornhole with Gemini 4 Argon ↗
AI Radar wrote this summary from the publisher’s notes with AI, not from a transcript. How it works.