Update README.md
Browse files
README.md
CHANGED
|
@@ -0,0 +1,54 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-nc-4.0
|
| 3 |
+
---
|
| 4 |
+
|
| 5 |
+
## Simulation GR1 Tabletop Task 1K Dataset
|
| 6 |
+
|
| 7 |
+
## Dataset Description:
|
| 8 |
+
The PhysicalAI-Robotics-GR00T-Teleop-GR1 dataset consists of 1000 teleoperation trajectories in simulation using the GR1 robot with upper body control. The simulation setup mimics tabletop manipulation tasks and uses RGB observations with a virtual camera. The robot is equipped with simulated Fourier hands.
|
| 9 |
+
|
| 10 |
+
This dataset is ready for non-commercial use.
|
| 11 |
+
|
| 12 |
+
## Dataset Owner(s):
|
| 13 |
+
NVIDIA GEAR
|
| 14 |
+
|
| 15 |
+
## Dataset Creation Date:
|
| 16 |
+
June 1, 2025
|
| 17 |
+
|
| 18 |
+
## License/Terms of Use:
|
| 19 |
+
This dataset is licensed under the [Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0)](https://creativecommons.org/licenses/by-nc/4.0/legalcode).
|
| 20 |
+
|
| 21 |
+
## Intended Usage:
|
| 22 |
+
Researchers, Academics, Open-Source Community: AI-driven robotics research and algorithm development in simulation environments.
|
| 23 |
+
Developers: Integrate and customize AI policies for tabletop tasks.
|
| 24 |
+
Startups & Companies: Non-commercial benchmarking and prototyping.
|
| 25 |
+
|
| 26 |
+
## Dataset Characterization
|
| 27 |
+
**Data Collection Method**
|
| 28 |
+
[Human Teleoperation in Simulation]
|
| 29 |
+
|
| 30 |
+
**Labeling Method**
|
| 31 |
+
[Human]
|
| 32 |
+
|
| 33 |
+
## Dataset Format
|
| 34 |
+
HDF5 format and LeRobot-style dataset
|
| 35 |
+
|
| 36 |
+
Modality:
|
| 37 |
+
- Observation: 44 dim of vectorized state (joint positions of upper body + waist)
|
| 38 |
+
- Action: 44 dim of vectorized action (joint positions of upper body + waist)
|
| 39 |
+
- Video: RGB video, 256x256 resolution (cropped and padded; original resolution is 51x320), 20 fps
|
| 40 |
+
- Language Instruction Example:
|
| 41 |
+
- *"pick up the bottled water, place it into the cabinet and close the cabinet"*
|
| 42 |
+
- *"pick the cupcake from the cutting board and place it in the basket"*
|
| 43 |
+
|
| 44 |
+
## Dataset Quantification
|
| 45 |
+
- 1000 teleoperation trajectories for each of the 24 tabletop tasks
|
| 46 |
+
- Total data storage (HDF5 format): ~14GB
|
| 47 |
+
- Total data storage (LeRobot-style dataset): ~39GB
|
| 48 |
+
|
| 49 |
+
## Download the dataset
|
| 50 |
+
|
| 51 |
+
```bash
|
| 52 |
+
huggingface-cli download \
|
| 53 |
+
--repo-type dataset nvidia/PhysicalAI-Robotics-GR00T-Teleop-GR1 \
|
| 54 |
+
--local-dir ./datasets/
|