Saved in:
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2505.15880 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908375990665216 |
|---|---|
| author | Xu, Zhiyuan Li, Bohan Gao, Huan-ang Gao, Mingju Chen, Yong Liu, Ming Yan, Chenxu Zhao, Hang Feng, Shuo Zhao, Hao |
| author_facet | Xu, Zhiyuan Li, Bohan Gao, Huan-ang Gao, Mingju Chen, Yong Liu, Ming Yan, Chenxu Zhao, Hang Feng, Shuo Zhao, Hao |
| contents | Generating photorealistic driving videos has seen significant progress recently, but current methods largely focus on ordinary, non-adversarial scenarios. Meanwhile, efforts to generate adversarial driving scenarios often operate on abstract trajectory or BEV representations, falling short of delivering realistic sensor data that can truly stress-test autonomous driving (AD) systems. In this work, we introduce Challenger, a framework that produces physically plausible yet photorealistic adversarial driving videos. Generating such videos poses a fundamental challenge: it requires jointly optimizing over the space of traffic interactions and high-fidelity sensor observations. Challenger makes this affordable through two techniques: (1) a physics-aware multi-round trajectory refinement process that narrows down candidate adversarial maneuvers, and (2) a tailored trajectory scoring function that encourages realistic yet adversarial behavior while maintaining compatibility with downstream video synthesis. As tested on the nuScenes dataset, Challenger generates a diverse range of aggressive driving scenarios-including cut-ins, sudden lane changes, tailgating, and blind spot intrusions-and renders them into multiview photorealistic videos. Extensive evaluations show that these scenarios significantly increase the collision rate of state-of-the-art end-to-end AD models (UniAD, VAD, SparseDrive, and DiffusionDrive), and importantly, adversarial behaviors discovered for one model often transfer to others. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_15880 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Challenger: Affordable Adversarial Driving Video Generation Xu, Zhiyuan Li, Bohan Gao, Huan-ang Gao, Mingju Chen, Yong Liu, Ming Yan, Chenxu Zhao, Hang Feng, Shuo Zhao, Hao Computer Vision and Pattern Recognition Generating photorealistic driving videos has seen significant progress recently, but current methods largely focus on ordinary, non-adversarial scenarios. Meanwhile, efforts to generate adversarial driving scenarios often operate on abstract trajectory or BEV representations, falling short of delivering realistic sensor data that can truly stress-test autonomous driving (AD) systems. In this work, we introduce Challenger, a framework that produces physically plausible yet photorealistic adversarial driving videos. Generating such videos poses a fundamental challenge: it requires jointly optimizing over the space of traffic interactions and high-fidelity sensor observations. Challenger makes this affordable through two techniques: (1) a physics-aware multi-round trajectory refinement process that narrows down candidate adversarial maneuvers, and (2) a tailored trajectory scoring function that encourages realistic yet adversarial behavior while maintaining compatibility with downstream video synthesis. As tested on the nuScenes dataset, Challenger generates a diverse range of aggressive driving scenarios-including cut-ins, sudden lane changes, tailgating, and blind spot intrusions-and renders them into multiview photorealistic videos. Extensive evaluations show that these scenarios significantly increase the collision rate of state-of-the-art end-to-end AD models (UniAD, VAD, SparseDrive, and DiffusionDrive), and importantly, adversarial behaviors discovered for one model often transfer to others. |
| title | Challenger: Affordable Adversarial Driving Video Generation |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2505.15880 |