Decoupled Diffusion Sparks Adaptive Scene Generation
January 6, 2026 · View on GitHub
Important
🌟 Stay up to date at opendrivelab.com!
Yunsong Zhou, Naisheng Ye, William Ljungbergh, Tianyu Li, et al.
- :mailbox_with_mail: Primary contact: Yunsong Zhou ( zhouyunsong2017@gmail.com )
Highlights
:fire: Nexus is a noise-decoupled prediction pipeline designed for adaptive driving scene generation, ensuring both timely reaction⏲️ and goal-directed control🥅.
:star2: Nexus can generate realistic safety-critical driving scenarios by flexibly controlling the future state of a scene, with the assistance of NeRF.
News
[2024/04]Nexus paper released.[2025/04]Nexus code and data initially released.
Table of Contents
TODO List
- Guidance tutorial
- Training code
- Nexus & checkpoint
- Initial repo & paper
Getting Started
Dataset
Nexus-Data is induced from real-world scenarios, in which we can obtain real-world map topology and layout. It also includes hazardous driving behaviors through interactions introduced by adversarial traffic generation. The safety-critical scenarios (on nuPlan dataset) can be obtained through this 🔗data link.
License and Citation
All assets and code in this repository are under the Apache 2.0 license unless specified otherwise. The data is under CC BY-NC-SA 4.0. Please consider citing our paper and project if they help your research.
@inproceedings{zhou2025nexus,
title={Decoupled Diffusion Sparks Adaptive Scene Generation},
author={Zhou, Yunsong and Ye, Naisheng and Ljungbergh, William and Li, Tianyu and Yang, Jiazhi and Yang, Zetong and Zhu, Hongzi and Petersson, Christoffer and Li, Hongyang},
booktitle={ICCV},
year={2025}
}
Related Resources
We acknowledge all the open-source contributors for the following projects to make this work possible:
