计算机科学 ›› 2025, Vol. 52 ›› Issue (10): 275-286.doi: 10.11896/jsjkx.240800030

• 人工智能 • 上一篇    下一篇

GCE3S:基于进化搜索的自动驾驶安全关键场景生成方法

孙乐乐, 黄松, 郑长友, 夏春艳, 阳真   

  1. 中国人民解放军陆军工程大学指挥控制工程学院 南京 210007
  • 收稿日期:2024-08-05 修回日期:2024-11-26 出版日期:2025-10-15 发布日期:2025-10-14
  • 通讯作者: 黄松(hs0317@163.com)
  • 作者简介:(sunlele@aeu.edu.cn)

GCE3S:A Method for Generating Safety-critical Scenarios in Autonomous Driving Based on Evolutionary Search

SUN Lele, HUANG Song, ZHENG Changyou, XIA Chunyan, YANG Zhen   

  1. College of Command and Control Engineering,Army Engineering University of PLA,Nanjing 210007,China
  • Received:2024-08-05 Revised:2024-11-26 Online:2025-10-15 Published:2025-10-14
  • About author:SUN Lele,born in 1999,postgraduate,is a member of CCF(No.V0414G).His main research interests include software testing and automous driving testing.
    HUANG Song,born in 1970,Ph.D,professor,Ph.D supervisor, is a distinguished member of CCF(No.29597S).His main research interests include software engineering,software security,software testing and quality assessment.

摘要: 自动驾驶技术的快速发展为交通出行带来了巨大的变革潜力,但在现实交通环境中,自动驾驶车辆的安全违规行为会导致巨大的损失。为了确保自动驾驶系统能够在各种复杂的交通环境中安全运行,在部署到实际道路之前必须对其进行充分的测试。由于自动驾驶测试场景空间的复杂性和高维性,现有安全关键场景生成方法存在成本高昂、效率低等问题。因此,提出了一种基于进化搜索的自动驾驶安全关键场景生成方法——GCE3S。GCE3S将场景中的障碍物及其属性映射为基因组成的染色体结构,从而对障碍物(车辆、天气、行人等)进行更加细致的扰动,构建具有对抗性的安全关键场景,并通过多个目标函数引导进化搜索算法生成多样化的安全关键场景。此外,在工业级自动驾驶系统百度Apollo和LGSVL模拟环境中对GCE3S进行了对比实验,实验结果表明,在相同的时间内,GCE3S生成的安全关键场景数量相较于最好的基准MOSAT方法提升了20.4%,生成的安全关键场景在多样性上增加了20%。

关键词: 自动驾驶, 仿真测试, 测试场景, 安全关键场景生成, 多目标进化搜索

Abstract: The rapid development of automated driving technology has brought great potential for transforming mobility,but automated driving technology,as safety-critical software,will lead to huge losses due to safety violations of self-driving vehicles in real traffic environments.In order to ensure that autonomous driving systems can operate safely in various complex traffic environments,autonomous driving systems must be fully tested before being deployed on real roads.Due to the complexity and high dimensionality of the autonomous driving test scenario space,existing safety critical scenario generation methods suffer from high cost and low efficiency.Therefore,this paper proposes an evolutionary search-based safety-critical scenario generation method for autonomous driving-GCE3S.GCE3S constructs safety-critical scenarios with adversarial nature by mapping the obstacles and their attributes in the scenario to the chromosome structure of genetic composition,thus perturbing the obstacles(vehicles,weather,pedestrians,etc.) in a more detailed manner and guiding the evolutionary search algorithms through multiple objective functions to generate diverse safety critical scenarios.In addition,the GCE3S is experimentally evaluated in the simulated environments of Baidu Apollo,an industrial-grade autonomous driving system,and LGSVL.The experimental results show that the number of safety-critical scenarios generated by GCE3S improve by 20.4% and the generated safety-critical scenarios increase by 20% in terms of diversity in the same amount of time as compared to the best baseline MOSAT method.

Key words: Autonomous driving,Simulation testing,Test scenario,Safety critical scenario generation,Multi-objective evolutio-nary search

中图分类号: 

  • TP311
[1]Ministry of Industry and Information Technology,NationalStandards Commission.Notice of the two departments on the issuance of the Guidelines for the Construction of National Telematics Industry Standard System(Intelligent Networked Vehicles)(Version 2023) [EB/OL].[2024-07-09].https://www.gov.cn/zhengce/zhengceku/202307/content_6894735.htm.
[2]TESLA-Tesla China.Autopilot [EB/OL].[2024-07-09].ht-tps://www.tesla.cn/autopilot?isappinstalled=0.
[3]WAYMO.Waymo Driver[EB/OL].[2024-07-09].https://waymo.com/.
[4]HUAWEI.Qian Kun ADS 3.0 [EB/OL].[2024-07-09].https://auto.huawei.com/cn/.
[5]BAIDU.Apollo[EB/OL].[2024-05-19].https://apollo.baidu.com.
[6]JIANG Z M,DANG S B,LI H Y,et al.A review of research progress on scenario testing of self-driving cars[J].Automotive Technology,2022(8):10-22.
[7]ZHAO X M National Key R&D Programme(2021YFB2501200) Team,ZHAO X M.Research progress of autonomous driving test and evaluation technology[J].Journal of Transportation Engineering,2023,23:10-77.
[8]ZAMPETTI F,KAPUR R,PENTA M D,et al.An empiricalcharacterization of software bugs in open-source Cyber-Physical Systems[J].Journal of Systems and Software,2022,192:111425.
[9]CHEN J Q,SHU X X,LAN F C,et al.Construction of autonomous driving test scenarios with typical hazardous accident characteristics[J].Journal of South China University of Technology(Natural Science Edition),2021,49(5):1-8.
[10]GAMBI A,HUYNH T,FRASER G.Generating effective test cases for self-driving cars from police reports[C]//Proceedings of the ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering.2019:257-267.
[11]ZHANG X,CAI Y.Building Critical Testing Scenarios for Autonomous Driving from Real Accidents[C]//Proceedings of the 32nd ACM SIGSOFT International Symposium on Software Testing and Analysis.2023:462-474.
[12]LEILABADI S H,SCHMIDT S.In-depth analysis of autonomous vehicle collisions in california[C]//International Confe-rence on Intelligent Transportation Systems.2019:889-893.
[13]ARCAINI P,ZHANG X Y,ISHIKAWA F.Less is More:Simplification of Test Scenarios for Autonomous Driving System Testing[C]//15th IEEE Conference on Software Testing,Verification and Validation,ICST 2022.IEEE,2022:279-290.
[14]LI G,LI Y,JHA S,et al.AV-FUZZER:Finding Safety Viola-tions in Autonomous Driving Systems[C]//31st IEEE International Symposium on Software Reliability Engineering,ISSRE 2020.2020:25-36.
[15]FENG S,FENG Y H,YU C,et al.Testing Scenario LibraryGeneration for Connected and Automated Vehicles,Part I:Methodology[J].IEEE Transactionson Intelligent Transportation Systems,2021,22(3):1573-1582.
[16]FENG S,FENG Y,SUN H W,et al.Testing Scenario Library Generation for Connected and Automated Vehicles,Part II:Case Studies[J].IEEE Transaction on Intelligent Transportation Systems,2020,22(9):5635-5647.
[17]KIM S,LIU M,RHEE J J,et al.DriveFuzz:Discovering Autonomous Driving Bugs through Driving Quality-Guided Fuzzing[C]//Proceedings of the 2022 ACM SIGSAC Conference on Computer and Communications Security,CCS 2022.2022:1753-1767.
[18]LUO Y,ZHANG X Y,ARCAINI P,et al.Targeting Requirements Violations of Autonomous Driving Systems by Dynamic Evolutionary Search[C]//36th IEEE/ACM International Conference on Automated Software Engineering,ASE 2021.2021:279-291.
[19]ZHANG X,ZHAO W,SUN Y,et al.Testing Automated Driving Systems by Breaking Many Laws Efficiently[C]//Procee-dings of the 32nd ACM SIGSOFT International Symposium on Software Testing and Analysis,ISSTA 2023.2023:942-953.
[20]TIAN H X,JIANG Y,WU G Q,et al.MOSAT:finding safety violations of autonomous driving systems using multi-objectivegenetic algorithm[C]//Proceedings of the 30th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering,ESEC/FSE 2022.2022:94-106.
[21]HUAI Y Q,ALMANEE S,CHEN Y T Y,et al.scenoRITA:Generating Diverse,Fully Mutable,Test Scenarios for Autonomous Vehicle Planning[J].IEEE Transactions Software Engineering,2023,49(10):4656-4676.
[22]Ministry of Industry and Information Technology.Public Notice for Submission and Approval of Recommended National Standardsfor Automated Vehicle Driving Classification [EB/OL].[2024-07-09].https://www.miit.gov.cn/zwgk/wjgs/art/2020/art_9a7eb2afbd5c411e88b5bbfc7012d7b1.html.
[23]Ministry of Industry and Information Technology.Automotivedriving automation classification [EB/OL].[2024-07-09].https://std.samr.gov.cn/gb/search/gbDetailed?id=CA6C0E542CB4C983E05397BE0A0AED11.
[24]Ministry of Industry and Information Technology,Ministry ofPublic Security,Ministry of Housing and Urban-Rural Development,et al.Circular of four ministries on the pilot work of access and on-road passage of intelligent networked vehicles [EB/OL].[2024-07-09].https://www.gov.cn/zhengce/zhengceku/202311/content_6915788.htm.
[25]CCTV.China carries out a pilot project on access and on-road passage of intelligent networked vehicles [EB/OL].[2024-07-09].https://www.gov.cn/yaowen/shipin/202406/content_6955492.htm.
[26]DAI J R,LI Z R,ZHANG W Q,et al.Simulation fuzzy testing for unmanned systems:current status,challenges and perspectives [J].Computer Research and Development,2023,60(7):1433-1447.
[27]ZHU X L,WANG H C,YOU H M,et al.Review of autonomous driving intelligent system testing research[J].Journal of Software,2021,32(7):2056-2077.
[28]LG.SVL[EB/OL].[2024-07-13].https://svlsimulator.com/.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!