Nvidia's model GR00T 2 watches humans on video, then imagines itself as the robot

Nvidia released GR00T 2, an AI model that learns robot tasks by watching human hand movements in videos, trained on 400,000 hours of footage. The system shows variable success rates between 10 and 80 percent depending on the task.
Why it matters
This approach makes it practical to teach robots new skills without manual programming, accelerating robot deployment in real-world settings.
More on:Nvidia


