Teaching a humanoid robot to play cornhole with Gemini 4 Argon

In an internal test, Google DeepMind robotics researchers instructed Gemini 4 Argon to watch a video of a human playing cornhole. Using its video understanding and advanced reasoning capabilities, Gemini 4 built a virtual simulation in MuJoCo and accurately calculated the physics and whole-body control required for Apptronik’s Apollo 2 humanoid to sink seven consecutive shots in the real world. This Video-to-Sim-to-Real pipeline compresses weeks of manual modeling into a few hours, transforming ordinary videos into an infinite training ground for physical AI. Learn more about Gemini 4 Argon: blog.google/innovation-and-ai/models-and-research/gemini-mod ___ Subscribe to our channel youtube.com/@googledeepmind Find us on X x.com/GoogleDeepMind Follow us on Instagram instagram.com/googledeepmind Add us on Linkedin linkedin.com/company/deepmind/