Hwsss/FlatLab-Asset

FlatLab Asset Library

Contents

Category	Assets	Category	Assets	Category	Assets
Board	6	Book	8	Box	8
Chessboard	4	Clock	4	Cutting Board	6
Disk	6	Disk Case	4	Fabric	4
Keyboard	6	Laptop	2	Magazine	6
Mouse Pad	4	Pad	6	Painting	4
Photo Album	4	Plate	6	Scarf	4
Shorts	4	Skirt	4	Towel	4

Total: 104 assets in 21 categories.

Asset visualization

Repository layout

Flat-Object-Assets.zip
Flat-Object-Assets/
├── Board/
│   ├── Board1/
│   │   ├── .metadata/
│   │   ├── simready_usd/
│   │   └── com.nvidia.simready.packaging.json
│   └── ...
├── Book/
└── ...

Flat-Object-Assets/ contains 21 semantic category folders and 104 packaged assets. Each asset package contains its SimReady metadata, its entry-point USD under simready_usd/, and all associated dependencies. Preserve the directory hierarchy when downloading or relocating the assets because USD references may be relative.

Flat-Object-Assets.zip provides the same asset collection as a single downloadable archive.

Download

hf download Hwsss/FlatLab-Asset \
  --repo-type dataset \
  --local-dir FlatLab-Asset

Usage

Open an entry-point file such as Flat-Object-Assets/Book/Book1/simready_usd/Book1.usd in a compatible USD application or reference it from an NVIDIA Isaac Sim scene. On the Hugging Face Hub, click an entry-point .usd file to open the OpenUSD Viewer. Some objects use PhysX collision/contact simulation and some deformable objects require Position-Based Dynamics support.

Project

• Project page: https://flatlab-web.github.io/
• Paper: FlatLab: A Unified Methodology Framework and Simulation-Based Benchmark for Robotic Manipulation of Flat Objects

License

The FlatLab asset release is provided under the Creative Commons Attribution 4.0 International license. See LICENSE for details. Bundled third-party components, if any, remain subject to their original terms; see THIRD_PARTY_LICENSES.md.

Citation

@inproceedings{zhu2026flatlab,
title={FlatLab: A Unified Methodology Framework and Simulation-Based Benchmark for Robotic Manipulation of Flat Objects},
author={Xingyu Zhu and Wenshuo Han and Zhouyu Wang and Yuran Wang and Ruihai Wu and Hao Dong and Fan Tang and Hechang Chen and Hyung Jin Chang and Yixing Gao},
booktitle={Forty-third International Conference on Machine Learning},
year={2026},
url={https://openreview.net/forum?id=1dWG9PJSVp}
}