• All Robots in One: A New Standard and Unified Dataset for Versatile, General-Purpose Embodied Agents
• • • 1. ARIO Overview
    • 2. Statistics of the ARIO dataset
    • 3. Comparison between ARIO dataset and other open source embodied datasets
    • 4. ARIO Dataset Format
    • 5. Download Links (data uploading and continuous updating)
    • • Transformation from open source datasets
      • Simulation data
      • Real collection data
      • Tips: After jumping to the sub dataset page, you should go to the dataset part to download the data, see the example below.
    • 6. Citation
    • 7. Contact
    • 8. License
    • 9. Acknowledgements

All Robots in One: A New Standard and Unified Dataset for Versatile, General-Purpose Embodied Agents

[Project page]   [Paper]

Zhiqiang Wang²*, Hao Zheng²*, Yunshuang Nie³*, Wenjun Xu¹*, Qingwei Wang²*, Hua Ye¹*, Zhe Li²,Kaidong Zhang³, Xuewen Cheng¹, Wanxi Dong², Chang Cai¹, Liang Lin^1,3, Feng Zheng^1,2†, Xiaodan Liang^1,3†

¹Pengcheng Laboratory   ²Southern University of Science and Technology   ³Sun Yat-sen University

*Equal contribution   †Corresponding Author

---

1. ARIO Overview

2. Statistics of the ARIO dataset

3. Comparison between ARIO dataset and other open source embodied datasets

4. ARIO Dataset Format

For the detailed format of the ARIO dataset, please refer to the ARIO Dataset Format.

5. Download Links (data uploading and continuous updating)

• Transformation from open source datasets

ManiWAV, Stanford UMI manipulation task, with sound data, 5 tasks, 1297 episodes
RH20T, Shanghai Jiao Tong University, 12719 episodes, multi tasks, 2 dataset files, one with tactile
Open-X-Embodiment, 2327939 episodes, multi tasks, multi agents

• Simulation data

Isaac_Sim, Isaac sim, multi agents and tasks
MuJoCo_UR5, MuJoCo simulation and UR5 robot model, a simple example of ARIO dataset format
MuJoCo_UR5_Opendrawers, MuJoCo simulation and UR5 robot model, open big and small drawers on table, 3 tasks * 50 episodes * 2 datafiles
MuJoCo_UR5_PickPlace, MuJoCo simulation and UR5 robot model, pick 14 objects on a table with several interfering objects and place into one of 3 baskets, 14 tasks * 50 episodes
MuJoCo_UR5_Grasp, MuJoCo simulation and UR5 robot model, grasp 14 objects on a table with several interfering objects, 14 tasks * 50 episodes
SeaWave, Dataa SeaWave platform, Cloud Ginger XR-1 robot, has various robotic manipulation skills, such as pick, place, and move near, in different task scenarios
HM3D Object Navigation, Habitat simulator, HM3D indoor dataset and the train split of HM3D object navigation dataset of Habitat Challenge 2022, 664241 episodes

• Real collection data

Songling_datasets, Songling Cobot Magic hardware platform, 2 tasks, 70 episodes, 3 RGBDs, 2 arms
SeaWave, Dataa Robot’s Cloud Ginger XR-1, 3 tasks and about 800 episodes.
PCL_CollectInReal, Cobot Magic hardware platform, 48 tasks, 2414 episodes, 3 RGBDs, 2 arms
...

Tips: After jumping to the sub dataset page, you should go to the dataset part to download the data, see the example below.

6. Citation

If you find ARIO dataset useful, feel free to cite our paper:

@misc{wang2024robotsonenewstandard,
      title={All Robots in One: A New Standard and Unified Dataset for Versatile, General-Purpose Embodied Agents}, 
      author={Zhiqiang Wang and Hao Zheng and Yunshuang Nie and Wenjun Xu and Qingwei Wang and Hua Ye and Zhe Li and Kaidong Zhang and Xuewen Cheng and Wanxi Dong and Chang Cai and Liang Lin and Feng Zheng and Xiaodan Liang},
      year={2024},
      eprint={2408.10899},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2408.10899}, 
}

7. Contact

If you have questions about the ARIO dataset, don't hesitate to reach out to CaiChang at caich@pcl.ac.cn.

8. License

For the real robot data collected by our team or materials generated by the simulation platform developed by us, are licensed under the Creative Commons Attribution 4.0 International License (CC-BY) or MIT. For materials converted from open source datasets or generated by simulation platform developed by others, we just follow their original license while publishing. For details, pay attention to the license of each open source project.

9. Acknowledgements

This work was supported by Research Institute of Multiple Agents and Embodied Intelligence, Peng Cheng Laboratory.
We extend our gratitude to the various open-source, datasets and platforms, including Open X-Embodiment, RH20T, ManiWAV, JD ManiData, and the contributors from Open X-Embodiment. Their contributions were vital in creating the ARIO dataset. Special thanks to the Habita-sim simulation platform, Habitat-lab module library, Habitat-Matterport 3D Dataset (HM3D) indoor dataset, and the Habitat Challenge organized by Facebook AI Research. It is through your open-source support that we were able to collect navigation simulation data. We are grateful for the Scaling Up and Distilling Down project for the simulation framework and the MuJoCo physics engine, aiding in generating simulation manipulation data. We appreciate the ARIO Embodied Intelligence Data Open Alliance members, such as Southern University of Science and Technology, Sun Yat-sen University, Dataa Robotics, Agilex Robotics, and JD Technology, for their technical support and contributions to the ARIO dataset development. The collaborative efforts have significantly advanced embodied AI research through the creation of the ARIO dataset.