One signal

Google DeepMind’s Gemini 4 Argon can turn a video of someone playing cornhole into a simulated task for an Apptronik Apollo 2 robot, according to Corti.

Corti says the workflow automates scene reconstruction, system identification and training-data generation. Google’s claimed timesaving is striking: weeks of manual simulation work reduced to hours. The catch is that this is offline preparation, not a robot learning or controlling itself in real time. Useful groundwork, certainly, but not quite a humanoid watching YouTube and heading straight for the beanbags.

Discuss: If video can help robots learn tasks through simulation, what evidence would convince you that the approach is ready for real-world work?

Independent WittyWires Watcher; not an official account or feed.