From Particles to Perils: SVGD-Based Hazardous Scenario Generation for Autonomous Driving Systems Testing
Proceedings of the ACM on Software Engineering · 30.06.2026 · 3 (FSE) · 3298-3320
DOI: 10.1145/3808153 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the ACM on Software Engineering, online erschienen am 30.06.2026. Verfasst von Linfeng Liang und 3 weiteren Autor:innen. Die Arbeit nennt 47 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: FT, DP.
Abstract (Originalsprache)
Simulation-based testing of autonomous driving systems (ADS) must uncover realistic and diverse failures in dense, heterogeneous traffic. However, existing search-based seeding methods (e.g., genetic algorithms) struggle in high-dimensional spaces, often collapsing to limited modes and missing many failure scenarios. We present PtoP, a framework that combines adaptive random seed generation with Stein Variational Gradient Descent (SVGD) to produce diverse, failure-inducing initial conditions. SVGD balances attraction toward high-risk regions and repulsion among particles, yielding risk-seeking yet well-distributed seeds across multiple failure modes. PtoP is plug-and-play and enhances existing online testing methods (e.g., reinforcement learning--based testers) by providing principled seeds. Evaluation in CARLA on two industry-grade ADS (Apollo, Autoware) and a native end-to-end system shows that PtoP improves safety violation rate (up to 27.68%), scenario diversity (9.6%), and map coverage (16.78%) over baselines.
Förderung
- FT
- DP
Zitieren
APA
Liang, L., Cheng, X., Chen, T. Y., & Zheng, X. (2026). From Particles to Perils: SVGD-Based Hazardous Scenario Generation for Autonomous Driving Systems Testing. Proceedings of the ACM on Software Engineering, 3(FSE), 3298-3320. https://doi.org/10.1145/3808153
Vancouver
Liang L, Cheng X, Chen TY, Zheng X. From Particles to Perils: SVGD-Based Hazardous Scenario Generation for Autonomous Driving Systems Testing. Proceedings of the ACM on Software Engineering. 2026;3(FSE):3298-3320. doi:10.1145/3808153
BibTeX
@article{Liang2026,
author = {Liang, Linfeng and Cheng, Xiao and Chen, Tsong Yueh and Zheng, Xi},
title = {From Particles to Perils: SVGD-Based Hazardous Scenario Generation for Autonomous Driving Systems Testing},
journal = {Proceedings of the ACM on Software Engineering},
year = {2026},
volume = {3},
number = {FSE},
pages = {3298--3320},
doi = {10.1145/3808153}
}
EUVIMED