Digest · past 24h

1. BeingBeyond Unveils Being-H0.8, the First Implicit Tactile World-Action Model Trained on 500,000 Hours of Video
   Chinese AI company BeingBeyond (智在无界) has released Being-H0.8, the first implicit tactile world-action model based on human video data. Trained on over 500,000 hours of first-person video, the model integrates tactile feedback into the prediction-execution-adjustment loop, enabling robots to anticipate contact and adjust actions in real time. The system successfully performed high-precision tasks like retrieving objects from a bag, writing with a brush, squeezing toothpaste, and picking up chips with a robotic arm.