README.md

---
license: cc
task_categories:
  - robotics
language:
  - en
tags:
  - robotics
  - robot-learning
  - bimanual-manipulation
  - imitation-learning
  - robotwin
  - robotwin2.0
pretty_name: RoboTwin 2.0
---

# RoboTwin 2.0 Dataset

Dexdata format RoboTwin 2.0 dataset.

## Download and Extract

If the dataset is uploaded as split tar archives:

```bash
cat robotwin2.tar.part-* | tar -xf -
```

After extraction, the dataset directory will be:

```text
robotwin2.0/
```

## Dataset Structure

```text
robotwin2.0/
├── jsonl/
│   └── <task_name>/
│       ├── clean/
│       └── randomized/
└── video/
    └── <task_name>/
        ├── clean/
        └── randomized/
```

Each task directory contains two data variants:

- `clean`: data generated in the clean/default setting
- `randomized`: data generated with domain randomization

## Citation

```bibtex
@article{chen2025robotwin,
  title={RoboTwin 2.0: A Scalable Data Generator and Benchmark with Strong Domain Randomization for Robust Bimanual Robotic Manipulation},
  author={Chen, Tianxing and Chen, Zanxin and Chen, Baijun and Cai, Zijian and Liu, Yibin and Liang, Qiwei and Li, Zixuan and Lin, Xianliang and Ge, Yiheng and Gu, Zhenyu and others},
  journal={arXiv preprint arXiv:2506.18088},
  year={2025}
}
```