1. Purpose of this Release

This repository contains a selected academic subset of the PAVE dataset, released to support non-commercial research, method development, and scientific publication.

The PAVE dataset is a real-world, end-to-end benchmark for evaluating production autonomous vehicles (AVs). It was collected entirely under identified autonomous driving mode, with synchronized sensors and high-precision localization. The dataset supports AV behavior and safety analysis, trajectory evaluation, and benchmark development.
https://arxiv.org/abs/2511.14185 

This academic subset is only a partial segment of the full dataset.

2. License and Usage Restrictions

By accessing or using this subset, you agree to the following terms:

Permitted Uses

Academic research

Non-commercial evaluation and method benchmarking

Course projects, theses, dissertations

Submission of scientific papers to journals and conferences

Prohibited Uses

Any commercial or industrial use
(e.g., product development, internal performance benchmarking, deployment)

Distribution, resale, or sharing of this data subset

Use in proprietary or commercial systems

Commercial use requires a separate licensing agreement.

3. Mandatory Citation (Formal Requirement)

If you use this dataset (even partially) in your research, publications, or reports, you must cite the PAVE dataset paper as a primary reference:

@inproceedings{li2026pave,
  title={PAVE: An end-to-end dataset for production autonomous vehicle evaluation},
  author={Li, Xiangyu and Wang, Chen and Liu, Yumao and He, Dengbo and Zhang, Jiahao and Ma, Ke},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={1010--1018},
  year={2026}
}


This citation requirement is binding and part of the dataset’s academic usage agreement.

4. Authorship and Attribution

You are not permitted to list the original PAVE dataset authors as co-authors solely due to dataset usage unless they contribute significantly to a specific research contribution.

Proper citation is required and sufficient for acknowledgment.

5. Disclaimer

This dataset subset is provided “as is” without warranty of any kind. The dataset authors and maintainers do not guarantee completeness, accuracy, or fitness for particular purposes, and assume no liability for any use outcomes.

6. About the Full Dataset

The publicly released subset represents only a limited academic sample.

For:

the full dataset

commercial licenses

industrial evaluation services

extended annotations or benchmarks

please contact the dataset maintainers at:
kema@hkust-gz.edu.cn

7. Ethical Use

Users should:

uphold data privacy and applicable regulations

avoid misuse or misinterpretation

uphold rigorous scientific and ethical standards

8. Acknowledgement and Community

We appreciate your interest and encourage you to star the repository and include the citation in any publications that utilize this dataset.


Unauthorized commercial use of this dataset will be pursued under applicable licensing agreements and legal frameworks. If your planned use case might be considered commercial, please contact us [kema@hkust-gz.edu.cn] before proceeding.