| 引用本文: | 田浩翔,吴国全,魏峻,郭安,韩星烁,陈伟,王伟,叶丹.基于路网建模的自动驾驶关键场景生成与自适应演化方法.软件学报,2026,37(4):1671-1689 |
| |
|
| |
|
|
| 本文已被：浏览 2052次 下载 2113次 |
码上扫一扫！ |
|
|
| 基于路网建模的自动驾驶关键场景生成与自适应演化方法 |
|
田浩翔1,2,3, 吴国全1,2,3, 魏峻1,2,3, 郭安4, 韩星烁5, 陈伟1,2,3, 王伟1,2,3, 叶丹1,2,3
|
|
1.中国科学院 软件研究所, 北京 100190;2.中国科学院大学, 北京 100049;3.计算机科学国家重点实验室(中国科学院 软件研究所), 北京 100190;4.南京大学, 江苏 南京 210093;5.南京航空航天大学, 江苏 南京 211106
|
|
| 摘要: |
| 自动驾驶系统的安全性对于自动驾驶汽车在现实世界中的实施非常重要. 因此, 自动驾驶系统在公开发布和部署之前必须进行充分的评估. 如何生成多样化的安全关键测试场景是自动驾驶系统测试的关键任务. 现有的自动驾驶系统关键场景生成方法, 包括再现现实世界的交通事故和基于搜索的关键场景生成. 然而, 前者由于自动驾驶与人类驾驶存在鸿沟, 现实世界交通事故大多无法发现自动驾驶系统的问题; 后者由于传统搜索算法的局限性, 发现的问题相似度较高. 此外, 由于测试场景的空间非常庞大, 二者生成关键场景的效率较低. 为了解决上述问题, 提出LEADE, 一种基于道路网络建模的自动驾驶系统安全关键场景生成和自适应演化方法. 具体来说, 它根据用户的测试需求构建抽象场景, 并通过道路网络建模生成具体场景. 然后, LEADE采用改进的自适应进化搜索来生成各种安全关键场景来测试自动驾驶系统. 在工业级全栈自动驾驶系统平台百度Apollo上实施和评估LEADE. 实验结果表明, LEADE可以有效和高效地生成安全关键场景, 并揭露Apollo的10种不同安全违背行为. 它通过识别同一道路上的4种新型安全关键场景, 优于两种最先进的基于搜索的自动驾驶系统测试技术. |
| 关键词: 自动驾驶系统 仿真测试 场景程序生成 |
| DOI：10.13328/j.cnki.jos.007471 |
| 分类号:TP311 |
| 基金项目:国家自然科学基金(62472412); 中国科学院软件研究所创新基金重大重点项目(ISCAS-ZD-202302); 中国科学院软件研究所基础研究项目(ISCAS-JCMS-202402) |
|
| Road-network-modeling-based Safety-critical Scenario Generation and Adaptive Evolution Approach for Autonomous Driving |
|
TIAN Hao-Xiang1,2,3, WU Guo-Quan1,2,3, WEI Jun1,2,3, GUO An4, HAN Xing-Shuo5, CHEN Wei1,2,3, WANG Wei1,2,3, YE Dan1,2,3
|
|
1.Institute of Software, Chinese Academy of Sciences, Beijing 100190, China;2.University of Chinese Academy of Sciences, Beijing 100049, China;3.State Key Laboratory of Computer Science (Institute of Software, Chinese Academy of Sciences), Beijing 100190, China;4.Nanjing University, Nanjing 210093, China;5.Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China
|
| Abstract: |
| The safety of autonomous driving systems (ADSs) is crucial for the implementation of autonomous vehicles (AVs). Therefore, ADSs must undergo thorough evaluation before being released and deployed publicly. Generating diverse, safety-critical test scenarios is a key task for ADS testing. Existing methods for generating ADS test scenarios include reproducing real-world traffic accidents or using search-based techniques. However, the accident-based scenario often fails to uncover safety violations in ADSs due to the gap between human driving and ADSs. The search-based approach tends to produce scenarios with high similarity because of the limitations of the search algorithm. To address these issues, this study proposes LEADE, a road network modeling-based safety-critical scenario generation and adaptive evolution method for ADSs. Specifically, it constructs abstract scenarios from user test requirements and generates concrete scenarios through road network modeling. LEADE then employs an improved adaptive evolutionary search to generate diverse safety-critical scenarios for testing the ADS. LEADE is implemented and evaluated on an industrial-grade full-stack ADS platform, Baidu Apollo. Experimental results demonstrate that LEADE can effectively and efficiently generate safety-critical scenarios and expose 10 diverse safety violations of Apollo. LEADE outperforms two state-of-the-art search-based ADS testing techniques by identifying 4 new types of safety-critical scenarios on the same roads. |
| Key words: autonomous driving system (ADS) simulation test scenario program generation |
|
|
|
|
