ManipBench: Benchmarking Vision-Language Models for Low-Level Robot Manipulation
Paper • 2505.09698 • Published
Datasets:
ArXiv:
License:
The Dataset Viewer has been disabled on this dataset.
ManipBench data (unofficial mirror)
This is an unofficial mirror of the ManipBench data, copied unchanged from the authors' public Google Drive, so it can be moved to offline machines. All credit belongs to the original authors. Please cite their paper:
Zhao et al., ManipBench: Benchmarking Vision-Language Models for Low-Level Robot Manipulation, CoRL 2025. Project page: https://manipbench.github.io · Paper: https://arxiv.org/abs/2505.09698
The data has no separate license file. The terms of the original release apply; the project website is CC BY-SA 4.0. If you are an author and want this mirror removed, please open a discussion and it will be taken down.
| File | Contents | Size |
|---|---|---|
simplified_manipbench_rr.zip |
Real-robot questions (Bridge/DROID, Q1+Q2), simplified format | 4.0 GB |
Dataset_ManipBench_ExData.zip |
Real-robot questions, raw format for the official pipeline | 3.2 GB |
VLM_evaluation_sim.zip |
Simulation questions | 192 MB |
Single Fabric Tests v1.zip |
Fabric manipulation questions | 6.5 GB |
SHA256SUMS holds the checksums of all four zips. Check them with sha256sum -c SHA256SUMS.
- Downloads last month
- 34