The Decoder· Maximilian Schreiner·· 7d agoAI score70
NVIDIA Releases GR00T 2 Model: Learning Human Actions Through Video, Reducing Remote Operation Needs in Robot Training
Nvidia's model GR00T 2 watches humans on video, then imagines itself as the robot
AI digest
NVIDIA has released the GR00T 2 model, which learns human actions through video and reduces the need for remote operations in robot training. The model is trained on a large amount of video data, capable of predicting the movement of robot arms and generating corresponding control commands.
Source:The Decoder · the-decoder.com