BeingBeyond's Being-M0.7 teaches humanoid robots to walk and grasp from human video
Overview
BeingBeyond released Being-M0.7, a latent world-action model that lets a humanoid robot walk and manipulate objects at the same time.
The model is pretrained on more than 10,000 hours of human egocentric video and motion data before it is trained on robot trajectories, according to Pandaily's report.
On the SIMPLE simulation benchmark, Being-M0.7 succeeded in 128 of 180 trials, or 71.1%. On a Unitree G1 humanoid, it completed 13 of 15 walk-and-grasp trials, tying GR00T-N1.6 on the same test.
Written by AI from the articles below · updated Oct 11, 11:01 PM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- PandailyNewsBeingBeyond's Being-M0.7 teaches humanoid robots to walk and grasp using human video
AIBeingBeyond released Being-M0.7, a latent world-action model that lets a humanoid robot walk and manipulate objects at the same time. The model is pretrained on more than 10,000 hours of human egocentric video and motion data before it sees robot trajectories. It succeeded in 128 of 180 trials, or 71.1%, on the SIMPLE simulation benchmark, and tied GR00T-N1.6 on a Unitree G1 humanoid with 13 of 15 successful walk-and-grasp trials.
Heat trend
Not enough continuous observations to show a trend yet.