README.md

---
license: mit
task_categories:
- robotics
- reinforcement-learning
language:
- en
pretty_name: VoLN-UAV Dataset
---

# VoLN-UAV Dataset

This Hugging Face dataset entry is intended for the navigation data used by VoLN-UAV. The environment assets are hosted separately so that users can fetch the simulator package and the trajectory data independently.

## Hugging Face Entries

- env: https://huggingface.co/datasets/Louj/VoLN-UAV-ENV
- dataset: https://huggingface.co/datasets/Louj/VoLN-UAV-Dataset

## Data Organization

The release package provides the benchmark inputs required by VoLN-UAV:

- scene-level Train/Validation/Test split manifests;
- route JSON files with RGB frame references and pose-derived state fields;
- benchmark metadata for scenes, episodes, and checksums;
- optional copied RGB frames under `source/frames/` when the package is built in `copy` mode.

## Usage

1. Download the dataset package and the `env` package.
2. Unzip the dataset package.
3. Set `source_root` in the benchmark config to the unzipped `source/` directory.
4. Run `python -m voln_uav.cli.build_benchmark --config <config.yaml>`.

The generated `manifest.json` contains the release summary and Hugging Face resource links.

## Recommended Citation

Please cite the VoLN-UAV paper and this dataset repository once the manuscript metadata is finalized.

## Download Layout

The dataset is uploaded as independent ZIP shards under `metadata/`, `train/`, `val/`, and `test/`.
Each shard is below 5 GB. Extract `metadata/VoLN-UAV-metadata.zip` first, then extract the
split shards you need into the same directory so that paths such as
`source/frames/<scene>/<trajectory>/<frame>.png` match the JSONL metadata.

The train, validation, and test splits are episode-disjoint. The test split is held out on a
separate scene, while train and validation may share scenes.

Use `SHA256SUMS.txt` to verify downloaded shards.