Evans Han retweetledi
PointWorld scales up training of an action-conditioned 3D world model that predicts env dynamics given a single RGB-D image and robot actions.
Our code is now open-sourced! It includes the training & evaluation pipeline, as well as the full data annotation pipeline for obtaining accurate depth, extrinsics, and 3D tracks of the DROID and BEHAVIOR datasets.
Code: github.com/NVlabs/PointWo…
Wenlong Huang@wenlong_huang
What if we can simulate an *interactive 3D world*, from a single image, in the wild, in real time? Introducing PointWorld-1B: a large pre-trained 3D world model that predicts env dynamics given RGB-D capture and robot actions. 🌐 point-world.github.io from @Stanford @nvidia
English
