3D Diffuser Actor
Marries 3D scene representations with diffusion policies: lifts multi-view features into 3D tokens and denoises end-effector trajectories, setting state of the art on RLBench and CALVIN.
- 81%
- 21K
- 20 Hz
- 3.2K
Benchmarks
Numbers as reported by the authors on the named evaluation suite.
| Evaluation | Suite | Score |
|---|---|---|
RLBench PerAct setup (18 tasks) | RLBench | 81.3% |
RLBench GNFactor setup | RLBench | 78.4% |
CALVIN long-horizon | CALVIN ABC->D | 3.4 |
Training
3D scene-token diffusion policy
180M
8x A100 80GB
400 accelerator-hours
45 h
3.2K
Relative 3D attention over scene tokens, DDPM over trajectories, keyframe supervision.
Depth quality dominates: noisy point clouds cost more than smaller models.
Linked datasets, robots & papers
Resolving related models, datasets and papers…