Efficient Safety Testing of Autonomous Vehicles via Adaptive Search over Crash-Derived Scenarios

Rui Zhou · 2025

Read Efficient Safety Testing of Autonomous Vehicles via Adaptive Search over Crash-Derived Scenarios by Rui Zhou (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-06575.txt.

Description

Rui Zhou wrote a 2025 paper on faster safety testing for self-driving cars. The hardest tests are rare crash-like scenes, and finding them by brute force is slow. Zhou draws test scenes from real crashes stored in a Chinese road safety database. Crash records were rebuilt to recover the moments before impact.

A Baidu Apollo driving system then faced those scenes in simulation. Zhou adds a search method that adapts as it goes, mixing broad moves with fine tuning. The goal is to find the scenes where the car fails, and find them fast. Trials showed the search sped up testing a great deal. Quicker tests mean more flaws found before cars reach public roads.

Catalog record

CREATORRui Zhou
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE37 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-06575

Browse the full library · About the Tower of Babel