metadata
license: apache-2.0
Download
Download and unpack the samples from either Hugging Face or ModelScope into examples/data/.
Hugging Face
pip install -U huggingface_hub
hf download D-Robotics/Uranus-Demo-Data \
--repo-type dataset \
--local-dir ./examples/data
ModelScope
pip install -U modelscope
modelscope download \
--dataset D-Robotics/Uranus-Demo-Data \
--local_dir ./examples/data
Each episode lands in ./examples/data/<episode_id>/ and can be passed directly to main.py --sample-dir.
Each sample is a self-contained "XML-environment" directory:
sample_dir/
├── meta.json # prompt, cameras, mjcf_path, end_effectors, skeleton, fps
├── temporal.json # step_qpos: [{state, robot2world_transform}, ...]
├── mjcf/
│ └── <robot>.xml # robot MJCF with baked-in camera calibration & mounts
├── ref_images/
│ └── <camera>.png # one reference image per camera
└── gt/
└── <camera>.mp4 # ground-truth video per camera
Your own robot data can be converted into this same format and fed to inference as well, as long as it provides the per-frame qpos, the robot MJCF with calibrated cameras, and one reference image per camera.
Dataset Licenses
The Apache-2.0 license declared for this repository applies only to the original code and materials released by this project. Samples derived from third-party datasets remain subject to their respective upstream licenses and terms of use.
| Source dataset | License | Source |
|---|---|---|
| AgiBot World 2026 | CC BY-NC-SA 4.0 | Dataset page |
| AgiBot World Beta | CC BY-NC-SA 4.0 | Dataset page |
| DROID | Apache License 2.0 | Dataset website |
| RoboChallenge Table30 | - | Dataset page |
| RoboChallenge Table30 v2 | - | Dataset page |