RMBench: Memory-Dependent Manipulation Benchmark

RMBench: Memory-Dependent Robotic Manipulation Benchmark with Insights into Policy Design. Under Review, PDF | arXiv | Website | Join our Community 🔥

Tianxing Chen*, Yuran Wang*, Mingleyang Li*, Yan Qin*, Hao Shi, Zixuan Li, Yifan Hu, Yingsheng Zhang, Kaixuan Wang, Yue Chen, Hongcheng Wang, Renjing Xu, Ruihai Wu, Yao Mu, Yaodong Yang, Hao Dong†, Ping Luo†

🧑🏻‍💻 RMBench Usage

This project is built upon RoboTwin 2.0, and you can seamlessly transfer your policy code between the two projects.

1. Installation

First, prepare a conda environment.

conda create -n RMBench python=3.10 -y
conda activate RMBench

RMBench Repo: https://github.com/RoboTwin-Platform/RMBench

git clone https://github.com/RoboTwin-Platform/RMBench.git

Then, run script/_install.sh to install basic conda envs and CuRobo:

bash script/_install.sh

2. Download Assets

To download the assets, run the following command. If you encounter any rate-limit issues, please log in to your Hugging Face account by running huggingface-cli login:

bash script/_download_assets.sh

3. Download Data

Please run the following command to download all data.

bash script/_download_data.sh

If you need to collect the data (we actually recommend downloading it directly)

In RMBench, we always use demo_clean setting.

Running the following command will first search for a random seed for the target collection quantity, and then replay the seed to collect data.

Please strictly follow our tutorial in RoboTwin 2.0 Doc - Collect Data.

bash collect_data.sh ${task_name} ${task_config} ${gpu_id}
# Example: bash collect_data.sh cover_blocks demo_clean 0

4. Run Policies

1. Mem-0 (ours): See Mem-0 Document
2. DP: See DP Document
3. ACT: See ACT Document
4. Pi 0.5: See Pi 0.5 Document
5. X-VLA: See X-VLA Document
6. Other Policies (Pi0, RDT, etc): See Document and See Folder
7. Configure your policy: See Tutorial Here

👍 Citations

If you find our work useful, please consider citing:

@misc{chen2026rmbenchmemorydependentroboticmanipulation,
      title={RMBench: Memory-Dependent Robotic Manipulation Benchmark with Insights into Policy Design}, 
      author={Tianxing Chen and Yuran Wang and Mingleyang Li and Yan Qin and Hao Shi and Zixuan Li and Yifan Hu and Yingsheng Zhang and Kaixuan Wang and Yue Chen and Hongcheng Wang and Renjing Xu and Ruihai Wu and Yao Mu and Yaodong Yang and Hao Dong and Ping Luo},
      year={2026},
      eprint={2603.01229},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2603.01229}, 
}

🏷️ License

This repository is released under the MIT license. See LICENSE for additional details.