We introduce 3D World Simulator, an action-conditioned multimodal world modeling framework that jointly predicts appearance, geometry, and tactile observations from a unified 3D representation.
The model predicts the next latent from the last `--n-history` latents and actions.
材料说明的限制
阅读于 2026-11-04
each task has specific set parameters in `tasks.py`, including specific voxel grid dimensions, workspace bounds, whether it has tactile sensing, among other information.