https://t.co/WSiF7EY4JN

Nvidia releases GR00T 2, using more human video to train robots

GR00T 2 is an open world-action model for robot arms, hands and humanoids. Nvidia pretrained it on 402,730 hours of human head-camera video, 10,184 hours of handheld-gripper demonstrations and 23,500 hours of teleoperated robot data. Ordinary human video lacks motor commands, which previously made it difficult for such models to use.

Nvidia releases GR00T 2 robot model

GR00T 2 is designed for robot arms, hands and humanoids. The account says its pretraining used 402,730 hours of human head-camera video, 10,184 hours of handheld-gripper demonstrations and 23,500 hours of teleoperated robot data, aiming to make use of human video that lacks motor commands.