Article(id=1196058110214320506, tenantId=1146029695717560320, journalId=1189621681917173762, issueId=1196058106951156162, articleNumber=null, orderNo=null, doi=10.19620/j.cnki.1000-3703.20240092, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=research-article, receivedDate=null, receivedDateStr=null, revisedDate=1713110400000, revisedDateStr=2024-04-15, acceptedDate=null, acceptedDateStr=null, onlineDate=1763092075159, onlineDateStr=2025-11-14, pubDate=1748016000000, pubDateStr=2025-05-24, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1763092075159, onlineIssueDateStr=2025-11-14, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1763092075159, creator=13701087609, updateTime=1763092075159, updator=13701087609, issue=Issue{id=1196058106951156162, tenantId=1146029695717560320, journalId=1189621681917173762, year='2025', volume='', issue='5', pageStart='1', pageEnd='62', issueExtLink='null', onlineDate='null', pubDate='null', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=null, createTime=1763092074382, creator=13701087609, updateTime=1763092350927, updator=13701087609, preIssue=null, nextIssue=null, ext={EN=IssueExt(id=1196059266915288024, tenantId=1146029695717560320, journalId=1189621681917173762, issueId=1196058106951156162, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1196059266915288025, tenantId=1146029695717560320, journalId=1189621681917173762, issueId=1196058106951156162, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null}, startPage=39, endPage=46, ext={EN=ArticleExt(id=1196058110424035710, articleId=1196058110214320506, tenantId=1146029695717560320, journalId=1189621681917173762, language=EN, title=Automatic Generation Method of Autonomous Driving Simulation Test Scenarios Based on Tree-Structured Parzen Estimator, columnId=null, journalTitle=Automobile Technology, columnName=null, runingTitle=null, highlight=null, articleAbstract=

In view of the exponential increase in the number of key scene scenarios generated in high-dimensional space, and the difficulty of traditional artificial construction or random search methods to balance coverage and efficiency, this paper proposes a search method based on single-objective Tree structure Parzen Estimator (TPE) and Multi-ObjectiveTree structure Parzen Estimator (MOTPE). A software-in-the-loop automated simulation testing framework is built by using the CARLA simulator. Taking weather elements as an example, the critical scenario generation effects of the different search algorithms are compared. The experimental results indicate that the TPE-based search method and the MOTPE-based method increase the number of key scenarios generated by 3.11 times and 2.06 times, respectively, compared to the random search method. The MOTPE method is 1.53 times better than TPE in terms of scenario quality. When combined with scenario automaed generation and testing frameworks, these methods effectively address the issue of exploding scenario numbers, allowing for the discovery of scenarios with high testing value.

, correspAuthors=null, authorNote=null, correspAuthorsNote=null, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, authorCompany=null, fund=null, authors=null, authorsList=Qin Qin, Zhisheng Yang, Daoxin Li, Zhiwei Shen, Xiaolin Cao), CN=ArticleExt(id=1196058349121876385, articleId=1196058110214320506, tenantId=1146029695717560320, journalId=1189621681917173762, language=CN, title=基于树结构Parzen估计器的自动驾驶仿真测试关键场景生成方法, columnId=0, journalTitle=汽车技术, columnName=, runingTitle=null, highlight=null, articleAbstract=

针对高维空间中生成关键场景场景数量呈指数增长,传统人工构造或随机搜索方法难以兼顾覆盖率与效率问题,提出基于单目标树结构Parzen估计器(TPE)和多目标树结构Parzen估计器(MOTPE)的搜索方法。通过CARLA模拟器搭建软件在环自动化仿真测试框架,以天气要素为例,对比不同搜索算法的关键场景生成效果。试验结果表明:基于TPE的搜索方法和基于MOTPE的方法相较于随机搜索方法生成的关键场景数量分别提高3.11倍和2.06倍,MOTPE方法的场景质量方面是TPE的1.53倍,配合场景自动化生成与测试框架,可有效解决场景数量爆炸问题,发现具有高测试价值的场景。

, correspAuthors=null, authorNote=null, correspAuthorsNote=
杨志胜(2001—),硕士,主要研究方向为自动驾驶、汽车电子、软件测试,
, copyrightStatement=null, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=zR2Cf8GIBeV6UDatedMoHA==, magXml=dsVvMlIXLSaiG27oJkeJKA==, pdfUrl=null, pdf=MOohcCvXrY/KnYkJLBUMqQ==, pdfFileSize=5948674, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=hd9NYZ/HKQwaMRZu01gWZg==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=wkzIEyeRaeFKUlLG9Xw2Jw==, mapNumber=null, authorCompany=null, fund=null, authors=null, authorsList=秦琴, 杨志胜, 李道鑫, 沈知玮, 曹晓琳)}, authors=[Author(id=1196558422670881163, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1196558422758961549, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, authorId=1196558422670881163, language=EN, stringName=Qin Qin, firstName=Qin, middleName=null, lastName=Qin, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 Shanghai Polytechnic University, School of Intelligent Manufacturing and Control Engineering, Shanghai 201209, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1196558422821876110, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, authorId=1196558422670881163, language=CN, stringName=秦琴, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 上海第二工业大学智能制造与控制工程学院,上海 201209, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1196558422335336831, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, xref=1, ext=[AuthorCompanyExt(id=1196558422343725440, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422335336831, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 Shanghai Polytechnic University, School of Intelligent Manufacturing and Control Engineering, Shanghai 201209), AuthorCompanyExt(id=1196558422352114049, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422335336831, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 上海第二工业大学智能制造与控制工程学院,上海 201209)])]), Author(id=1196558422893179280, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=821419302@qq.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1196558422993842578, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, authorId=1196558422893179280, language=EN, stringName=Zhisheng Yang, firstName=Zhisheng, middleName=null, lastName=Yang, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 Shanghai Polytechnic University, School of Intelligent Manufacturing and Control Engineering, Shanghai 201209, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1196558423077728659, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, authorId=1196558422893179280, language=CN, stringName=杨志胜, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 上海第二工业大学智能制造与控制工程学院,上海 201209, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1196558422335336831, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, xref=1, ext=[AuthorCompanyExt(id=1196558422343725440, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422335336831, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 Shanghai Polytechnic University, School of Intelligent Manufacturing and Control Engineering, Shanghai 201209), AuthorCompanyExt(id=1196558422352114049, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422335336831, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 上海第二工业大学智能制造与控制工程学院,上海 201209)])]), Author(id=1196558423136448917, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1196558423203557783, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, authorId=1196558423136448917, language=EN, stringName=Daoxin Li, firstName=Daoxin, middleName=null, lastName=Li, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 Shanghai Polytechnic University, School of Intelligent Manufacturing and Control Engineering, Shanghai 201209, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1196558423266472344, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, authorId=1196558423136448917, language=CN, stringName=李道鑫, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, address=1 上海第二工业大学智能制造与控制工程学院,上海 201209, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1196558422335336831, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, xref=1, ext=[AuthorCompanyExt(id=1196558422343725440, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422335336831, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 Shanghai Polytechnic University, School of Intelligent Manufacturing and Control Engineering, Shanghai 201209), AuthorCompanyExt(id=1196558422352114049, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422335336831, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 上海第二工业大学智能制造与控制工程学院,上海 201209)])]), Author(id=1196558423388107162, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, orderNo=3, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1196558423459410332, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, authorId=1196558423388107162, language=EN, stringName=Zhiwei Shen, firstName=Zhiwei, middleName=null, lastName=Shen, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2 School of Electrical Engineering and Telecommunications, University of New South Wales, Sydney 2052, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1196558423518130589, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, authorId=1196558423388107162, language=CN, stringName=沈知玮, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, address=2 新南威尔士大学电气工程与通信学院,澳大利亚 悉尼 2052, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1196558422410834306, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, xref=2, ext=[AuthorCompanyExt(id=1196558422419222915, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422410834306, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 School of Electrical Engineering and Telecommunications, University of New South Wales, Sydney 2052), AuthorCompanyExt(id=1196558422423417220, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422410834306, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 新南威尔士大学电气工程与通信学院,澳大利亚 悉尼 2052)])]), Author(id=1196558423581045151, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, orderNo=4, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1196558423673319841, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, authorId=1196558423581045151, language=EN, stringName=Xiaolin Cao, firstName=Xiaolin, middleName=null, lastName=Cao, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=3, address=3 College of Automotive Engineering, Jilin University, Changchun 130015, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1196558423761400226, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, authorId=1196558423581045151, language=CN, stringName=曹晓琳, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=3, address=3 吉林大学汽车工程学院,长春 130015, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1196558422549246342, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, xref=3, ext=[AuthorCompanyExt(id=1196558422557634951, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422549246342, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 College of Automotive Engineering, Jilin University, Changchun 130015), AuthorCompanyExt(id=1196558422566023560, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422549246342, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 吉林大学汽车工程学院,长春 130015)])])], keywords=[Keyword(id=1196558423908200867, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, orderNo=1, keyword=Autonomous driving), Keyword(id=1196558423992086948, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, orderNo=2, keyword=Scenario generation), Keyword(id=1196558424055001509, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, orderNo=3, keyword=Tree-structured Parzen Estimator (TPE)), Keyword(id=1196558424109527462, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, orderNo=4, keyword=Multi-Objective Tree-structured Parzen Estimator (MOTPE)), Keyword(id=1196558424226967975, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, orderNo=5, keyword=CARLA), Keyword(id=1196558424285688232, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, orderNo=1, keyword=自动驾驶), Keyword(id=1196558424352797097, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, orderNo=2, keyword=场景生成), Keyword(id=1196558424424100266, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, orderNo=3, keyword=TPE), Keyword(id=1196558424507986347, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, orderNo=4, keyword=MOTPE), Keyword(id=1196558424570900908, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, orderNo=5, keyword=CARLA)], refs=[Reference(id=1196558427804709319, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2025, volume=61, issue=1, pageStart=59, pageEnd=79, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=孙乐乐, 黄松, 郑长友, journalName=计算机工程与应用, refType=null, unstructuredReference=孙乐乐, 黄松, 郑长友, 等. 自动驾驶仿真测试场景生成技术研究进展[J]. 计算机工程与应用, 2025, 61(1): 59-79., articleTitle=自动驾驶仿真测试场景生成技术研究进展, refAbstract=null), Reference(id=1196558427867623880, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2025, volume=61, issue=1, pageStart=59, pageEnd=79, url=null, language=null, rfNumber=[1], rfOrder=1, authorNames=SUN L L, HUANG S, ZHENG C Y, journalName=Computer Engineering and Applications, refType=null, unstructuredReference=SUN L L, HUANG S, ZHENG C Y, et al. Research Progress on Autonomous Driving Simulation Test Scenario Generation Technology[J]. Computer Engineering and Applications, 2025, 61(1): 59-79., articleTitle=Research Progress on Autonomous Driving Simulation Test Scenario Generation Technology, refAbstract=null), Reference(id=1196558427955704265, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=8, pageStart=10, pageEnd=22, url=null, language=null, rfNumber=[2], rfOrder=2, authorNames=蒋拯民, 党少博, 李慧云, journalName=汽车技术, refType=null, unstructuredReference=蒋拯民, 党少博, 李慧云, 等. 自动驾驶汽车场景测试研究进展综述[J]. 汽车技术, 2022(8): 10-22., articleTitle=自动驾驶汽车场景测试研究进展综述, refAbstract=null), Reference(id=1196558428073144778, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=8, pageStart=10, pageEnd=22, url=null, language=null, rfNumber=[2], rfOrder=3, authorNames=JIANG Z M, DANG S B, LI H Y, journalName=Automobile Technology, refType=null, unstructuredReference=JIANG Z M, DANG S B, LI H Y, et al. A Survey on the Research Progress of Scenario-Based Testing for Autonomous Vehicles[J]. Automobile Technology, 2022(8): 10-22., articleTitle=A Survey on the Research Progress of Scenario-Based Testing for Autonomous Vehicles, refAbstract=null), Reference(id=1196558428140253643, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2022, volume=35, issue=1, pageStart=316, pageEnd=333, url=null, language=null, rfNumber=[3], rfOrder=4, authorNames=邓伟文, 李江坤, 任秉韬, journalName=中国公路学报, refType=null, unstructuredReference=邓伟文, 李江坤, 任秉韬, 等. 面向自动驾驶的仿真场景自动生成方法综述[J]. 中国公路学报, 2022, 35(1): 316-333., articleTitle=面向自动驾驶的仿真场景自动生成方法综述, refAbstract=null), Reference(id=1196558428224139724, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2022, volume=35, issue=1, pageStart=316, pageEnd=333, url=null, language=null, rfNumber=[3], rfOrder=5, authorNames=DENG W W, LI J K, REN B T, journalName=China Journal of Highway and Transport, refType=null, unstructuredReference=DENG W W, LI J K, REN B T, et al. A Survey on Automatic Simulation Scenario Generation Methods for Autonomous Driving[J]. China Journal of Highway and Transport, 2022, 35(1): 316-333., articleTitle=A Survey on Automatic Simulation Scenario Generation Methods for Autonomous Driving, refAbstract=null), Reference(id=1196558428362551757, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=null, pageStart=2620, pageEnd=2627, url=null, language=null, rfNumber=[4], rfOrder=6, authorNames=RAMAKRISHNA S, LUO B T, KUHN C B, journalName=2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC), refType=null, unstructuredReference=RAMAKRISHNA S, LUO B T, KUHN C B, et al. ANTI-CARLA:An Adversarial Testing Framework for Autonomous Vehicles in CARLA[C]// 2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC). Macau, China: IEEE, 2022: 2620-2627., articleTitle=ANTI-CARLA:An Adversarial Testing Framework for Autonomous Vehicles in CARLA[C], refAbstract=null), Reference(id=1196558428450632142, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=null, pageStart=107, pageEnd=116, url=null, language=null, rfNumber=[5], rfOrder=7, authorNames=RAMAKRISHNA S, LUO B T, BARVE Y, journalName=Fajardo, PR, refType=null, unstructuredReference=RAMAKRISHNA S, LUO B T, BARVE Y, et al. Risk-Aware Scene Sampling for Dynamic Assurance of Autonomous Systems[C]// 2022 IEEE International Conference on Assured Autonomy (ICAA). Fajardo, PR, USA: IEEE, 2022: 107-116., articleTitle=Risk-Aware Scene Sampling for Dynamic Assurance of Autonomous Systems[C]// 2022 IEEE International Conference on Assured Autonomy (ICAA), refAbstract=null), Reference(id=1196558428542906831, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2023, volume=40, issue=9, pageStart=2781, pageEnd=2784, url=null, language=null, rfNumber=[6], rfOrder=8, authorNames=秦琴, 谷文军, journalName=计算机应用研究, refType=null, unstructuredReference=秦琴, 谷文军. Ball Tree优化的自动驾驶仿真测试场景生成方法[J]. 计算机应用研究, 2023, 40(9): 2781-2784+2791., articleTitle=Ball Tree优化的自动驾驶仿真测试场景生成方法, refAbstract=null), Reference(id=1196558428614210000, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2023, volume=40, issue=9, pageStart=2781, pageEnd=2784, url=null, language=null, rfNumber=[6], rfOrder=9, authorNames=QIN Q, GU W J, journalName=Application Research of Computers, refType=null, unstructuredReference=QIN Q, GU W J. Ball Tree Optimized Automatic Driving Simulation Test Scenario Generation Method. Application Research of Computers, 2023, 40(9): 2781-2784+2791., articleTitle=Ball Tree Optimized Automatic Driving Simulation Test Scenario Generation Method, refAbstract=null), Reference(id=1196558428693901777, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2021, volume=49, issue=8, pageStart=1162, pageEnd=1169, url=null, language=null, rfNumber=[7], rfOrder=10, authorNames=邢星宇, 吴旭阳, 刘力豪, journalName=同济大学学报(自然科学版), refType=null, unstructuredReference=邢星宇, 吴旭阳, 刘力豪, 等. 基于目标优化的自动驾驶决策规划系统自动化测试方法[J]. 同济大学学报(自然科学版), 2021, 49(8): 1162-1169., articleTitle=基于目标优化的自动驾驶决策规划系统自动化测试方法, refAbstract=null), Reference(id=1196558428765204946, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2021, volume=49, issue=8, pageStart=1162, pageEnd=1169, url=null, language=null, rfNumber=[7], rfOrder=11, authorNames=XING X Y, WU X Y, LIU L H, journalName=Journal of Tongji University (Natural Science), refType=null, unstructuredReference=XING X Y, WU X Y, LIU L H, et al. Automatic Testing Method Based on Optimization Algorithms for the Decision and Planning System of Autonomous Vehicles[J]. Journal of Tongji University (Natural Science), 2021, 49(8): 1162-1169., articleTitle=Automatic Testing Method Based on Optimization Algorithms for the Decision and Planning System of Autonomous Vehicles, refAbstract=null), Reference(id=1196558428832313811, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2011, volume=null, issue=null, pageStart=2546, pageEnd=2554, url=null, language=null, rfNumber=[8], rfOrder=12, authorNames=BERGSTRA J, BARDENET R, BENGIO Y, journalName=Proceedings of the 25th International Conference on Neural Information Processing Systems. Granada Spain:PMLR, refType=null, unstructuredReference=BERGSTRA J, BARDENET R, BENGIO Y, et al. Algorithms for Hyper-Parameter Optimization[C]// Proceedings of the 25th International Conference on Neural Information Processing Systems. Granada Spain:PMLR, 2011: 2546-2554., articleTitle=Algorithms for Hyper-Parameter Optimization, refAbstract=null), Reference(id=1196558428899422676, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=73, pageStart=1209, pageEnd=1250, url=null, language=null, rfNumber=[9], rfOrder=13, authorNames=OZAKI Y, TANIGAKI Y, WATANABE S, journalName=Journal of Artificial Intelligence Research, refType=null, unstructuredReference=OZAKI Y, TANIGAKI Y, WATANABE S, et al. Multiobjective Tree-Structured Parzen Estimator[J]. Journal of Artificial Intelligence Research, 2022(73): 1209-1250., articleTitle=Multiobjective Tree-Structured Parzen Estimator, refAbstract=null), Reference(id=1196558428983308757, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2018, volume=null, issue=null, pageStart=1821, pageEnd=1827, url=null, language=null, rfNumber=[10], rfOrder=14, authorNames=MENZEL T, BAGSCHIK G, MAURER M, journalName=Changshu, refType=null, unstructuredReference=MENZEL T, BAGSCHIK G, MAURER M. Scenarios for Development,Test and Validation of Automated Vehicles[C]// 2018 IEEE Intelligent Vehicles Symposium (IV). Changshu, China: IEEE, 2018: 1821-1827., articleTitle=Scenarios for Development,Test and Validation of Automated Vehicles[C]// 2018 IEEE Intelligent Vehicles Symposium (IV), refAbstract=null), Reference(id=1196558429058806230, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2016, volume=24, issue=3, pageStart=521, pageEnd=544, url=null, language=null, rfNumber=[11], rfOrder=15, authorNames=GUERREIRO A P, FONSECA C M, PAQUETE L, journalName=Evolutionary Computation, refType=null, unstructuredReference=GUERREIRO A P, FONSECA C M, PAQUETE L. Greedy Hypervolume Subset Selection in Low Dimensions[J]. Evolutionary Computation, 2016, 24(3): 521-544., articleTitle=Greedy Hypervolume Subset Selection in Low Dimensions, refAbstract=null), Reference(id=1196558429138498007, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[12], rfOrder=16, authorNames=PIRRI P, PAHL C, IOINI N E, journalName=2021 IEEE 18th Annual Consumer Communications & Networking Conference (CCNC), refType=null, unstructuredReference=PIRRI P, PAHL C, IOINI N E, et al. Towards Cooperative Maneuvering Simulation:Tools and Architecture[C]// 2021 IEEE 18th Annual Consumer Communications & Networking Conference (CCNC). Las Vegas, NV, USA: IEEE, 2021., articleTitle=Towards Cooperative Maneuvering Simulation:Tools and Architecture[C], refAbstract=null), Reference(id=1196558429742477784, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2021, volume=43, issue=7, pageStart=1030, pageEnd=1036, url=null, language=null, rfNumber=[13], rfOrder=17, authorNames=白雪松, 邓伟文, 任秉韬, journalName=汽车工程, refType=null, unstructuredReference=白雪松, 邓伟文, 任秉韬, 等. 一种自动驾驶仿真场景要素的提取方法[J]. 汽车工程, 2021, 43(7): 1030-1036+1065., articleTitle=一种自动驾驶仿真场景要素的提取方法, refAbstract=null), Reference(id=1196558429809586649, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2021, volume=43, issue=7, pageStart=1030, pageEnd=1036, url=null, language=null, rfNumber=[13], rfOrder=18, authorNames=BAI X S, DENG W W, REN B T, journalName=Automotive Engineering, refType=null, unstructuredReference=BAI X S, DENG W W, REN B T, et al. An Extraction Method of Scenario Elements for Autonomous Driving Simulation[J]. Automotive Engineering, 2021, 43(7): 1030-1036+1065., articleTitle=An Extraction Method of Scenario Elements for Autonomous Driving Simulation, refAbstract=null), Reference(id=1196558429864112602, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2020, volume=null, issue=100, pageStart=66, pageEnd=75, url=null, language=null, rfNumber=[14], rfOrder=19, authorNames=CHEN D, ZHOU B, KOLTUN V, journalName=Proceedings of the Conference on Robot Learning. Auckland, refType=null, unstructuredReference=CHEN D, ZHOU B, KOLTUN V, et al. Learning by Cheating[C]// Proceedings of the Conference on Robot Learning. Auckland, New Zealand: PMLR, 2020(100): 66-75., articleTitle=Learning by Cheating, refAbstract=null), Reference(id=1196558429918638555, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2022, volume=null, issue=35, pageStart=25667, pageEnd=25682, url=null, language=null, rfNumber=[15], rfOrder=20, authorNames=XU C J, DING W H, LV W J, journalName=26th Conference on Neural Information Processing Systems Datasets and Benchmarks Track. New Orleans, refType=null, unstructuredReference=XU C J, DING W H, LV W J, et al. Safebench: A Benchmarking Platform for Safety Evaluation of Autonomous Vehicles[C]// 26th Conference on Neural Information Processing Systems Datasets and Benchmarks Track. New Orleans, USA: PMLR, 2022(35): 25667-25682., articleTitle=Safebench: A Benchmarking Platform for Safety Evaluation of Autonomous Vehicles, refAbstract=null), Reference(id=1196558430023496156, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, doi=null, pmid=null, pmcid=null, year=2020, volume=22, issue=3, pageStart=1573, pageEnd=1582, url=null, language=null, rfNumber=[16], rfOrder=21, authorNames=FENG S, FENG Y H, YU C H, journalName=IEEE Transactions on Intelligent Transportation Systems, refType=null, unstructuredReference=FENG S, FENG Y H, YU C H, et al. Testing Scenario Library Generation for Connected and Automated Vehicles, Part I: Methodology[J]. IEEE Transactions on Intelligent Transportation Systems, 2020, 22(3): 1573-1582., articleTitle=Testing Scenario Library Generation for Connected and Automated Vehicles, Part I: Methodology, refAbstract=null)], funds=null, companyList=[AuthorCompany(id=1196558422335336831, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, xref=1, ext=[AuthorCompanyExt(id=1196558422343725440, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422335336831, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 Shanghai Polytechnic University, School of Intelligent Manufacturing and Control Engineering, Shanghai 201209), AuthorCompanyExt(id=1196558422352114049, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422335336831, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1 上海第二工业大学智能制造与控制工程学院,上海 201209)]), AuthorCompany(id=1196558422410834306, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, xref=2, ext=[AuthorCompanyExt(id=1196558422419222915, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422410834306, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 School of Electrical Engineering and Telecommunications, University of New South Wales, Sydney 2052), AuthorCompanyExt(id=1196558422423417220, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422410834306, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2 新南威尔士大学电气工程与通信学院,澳大利亚 悉尼 2052)]), AuthorCompany(id=1196558422549246342, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, xref=3, ext=[AuthorCompanyExt(id=1196558422557634951, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422549246342, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 College of Automotive Engineering, Jilin University, Changchun 130015), AuthorCompanyExt(id=1196558422566023560, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, companyId=1196558422549246342, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3 吉林大学汽车工程学院,长春 130015)])], figs=[ArticleFig(id=1196558425741111725, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, label=null, caption=null, figureFileSmall=dFiQL+QirLVuRXCyYYyESg==, figureFileBig=lXNkpZswUjXFA7/DARIvJQ==, tableContent=null), ArticleFig(id=1196558425812414894, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, label=图1, caption=关键场景自动化生成与测试框架, figureFileSmall=dFiQL+QirLVuRXCyYYyESg==, figureFileBig=lXNkpZswUjXFA7/DARIvJQ==, tableContent=null), ArticleFig(id=1196558425904689583, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, label=null, caption=null, figureFileSmall=ZolJwXp88lJQXLSw7eE2ng==, figureFileBig=jpAnhAG7MJErIKxRA5Sypg==, tableContent=null), ArticleFig(id=1196558425971798448, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, label=图2, caption=图结构搜索空间, figureFileSmall=ZolJwXp88lJQXLSw7eE2ng==, figureFileBig=jpAnhAG7MJErIKxRA5Sypg==, tableContent=null), ArticleFig(id=1196558426030518705, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, label=null, caption=null, figureFileSmall=Mk0d9pK+9jiAqGA0f/Gnlg==, figureFileBig=b+zMN2OhJOvuF+j9fshiqQ==, tableContent=null), ArticleFig(id=1196558426089238962, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, label=图3, caption=yY*的关系示意, figureFileSmall=Mk0d9pK+9jiAqGA0f/Gnlg==, figureFileBig=b+zMN2OhJOvuF+j9fshiqQ==, tableContent=null), ArticleFig(id=1196558426147959219, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, label=null, caption=null, figureFileSmall=1mUmihIMWBhnZNqEIOpU4A==, figureFileBig=tk6minInc2FV8YRPdGNNYA==, tableContent=null), ArticleFig(id=1196558426202485172, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, label=图4, caption=软件在环仿真测试框架, figureFileSmall=1mUmihIMWBhnZNqEIOpU4A==, figureFileBig=tk6minInc2FV8YRPdGNNYA==, tableContent=null), ArticleFig(id=1196558426257011125, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, label=null, caption=null, figureFileSmall=GqzJ6fBfqshmoAJtMGZXkg==, figureFileBig=hDxCrGjzpSjys6/xQNFnjg==, tableContent=null), ArticleFig(id=1196558426328314294, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, label=图5, caption=CARLA仿真场景示意, figureFileSmall=GqzJ6fBfqshmoAJtMGZXkg==, figureFileBig=hDxCrGjzpSjys6/xQNFnjg==, tableContent=null), ArticleFig(id=1196558426391228855, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, label=null, caption=null, figureFileSmall=w47hQEGreVMSmrgMGgV3/g==, figureFileBig=J9H3NB83cS32LvF6/5Faiw==, tableContent=null), ArticleFig(id=1196558426441560504, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, label=图6, caption=小镇道路示意, figureFileSmall=w47hQEGreVMSmrgMGgV3/g==, figureFileBig=J9H3NB83cS32LvF6/5Faiw==, tableContent=null), ArticleFig(id=1196558426500280761, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, label=null, caption=null, figureFileSmall=mdzP8g/HjRNypQ4FsjBKqA==, figureFileBig=cl5GPr/Wj52BjytaPzUuYw==, tableContent=null), ArticleFig(id=1196558426571583930, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, label=图7, caption=不同方法统计评价指标结果, figureFileSmall=mdzP8g/HjRNypQ4FsjBKqA==, figureFileBig=cl5GPr/Wj52BjytaPzUuYw==, tableContent=null), ArticleFig(id=1196558426630304187, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, label=null, caption=null, figureFileSmall=d1cxSZHDjdWGbNUeAExh7A==, figureFileBig=B0xP+c6CvvUUtYtUK289VA==, tableContent=null), ArticleFig(id=1196558426693218748, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, label=图8, caption=不同方法搜索结果空间分布, figureFileSmall=d1cxSZHDjdWGbNUeAExh7A==, figureFileBig=B0xP+c6CvvUUtYtUK289VA==, tableContent=null), ArticleFig(id=1196558426751939005, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, label=null, caption=null, figureFileSmall=null, figureFileBig=null, tableContent=
输入:初始观测值D={(x(1),y(1)),…,(x(k),y(k))},迭代次数Nt,候选场景数Nc,分位数γ,约束条件R,约束条件数量Rn,逻辑场景参数的数量T
输出:关键场景集合。
for t =1,2,, Nt do
Dl={(x,y) ∈D | y包含在集合D中目标函数值小于y*的集合中}
Dg=D\Dl
S[ ]=NULL //下一轮测试具体场景列表
for i=1, 2,, T do //对所有逻辑场景参数进行采样
C[ ]=NULL //初始化候选参数列表
for j=1, 2,, Nc do
t=sample(l(xi)) //从l(xi)中采样参数值
for k=1, 2,, Rn do //检查所有约束
if 约束条件是关于逻辑场景参数i的then
(p,r)=getParea(i,t) //参数t所在节点的祖先节点的采样范围
if judge(p,r,S) //判断已经采样的父节点p是否在约束范围内
C[j]=t //将其赋值给候选参数值列表
j=j+1
end for
end for
S[i]= a r g m a x x i C il(xi)/g(xi) //将C中采集函数值最大的候选场景赋值给S[i]
end for
D=D∪{(S, f(S))}
end for
return D中关键场景集合H
), ArticleFig(id=1196558426823242174, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, label=表1, caption=

基于TPE的搜索算法

, figureFileSmall=null, figureFileBig=null, tableContent=
输入:初始观测值D={(x(1),y(1)),…,(x(k),y(k))},迭代次数Nt,候选场景数Nc,分位数γ,约束条件R,约束条件数量Rn,逻辑场景参数的数量T
输出:关键场景集合。
for t =1,2,, Nt do
Dl={(x,y) ∈D | y包含在集合D中目标函数值小于y*的集合中}
Dg=D\Dl
S[ ]=NULL //下一轮测试具体场景列表
for i=1, 2,, T do //对所有逻辑场景参数进行采样
C[ ]=NULL //初始化候选参数列表
for j=1, 2,, Nc do
t=sample(l(xi)) //从l(xi)中采样参数值
for k=1, 2,, Rn do //检查所有约束
if 约束条件是关于逻辑场景参数i的then
(p,r)=getParea(i,t) //参数t所在节点的祖先节点的采样范围
if judge(p,r,S) //判断已经采样的父节点p是否在约束范围内
C[j]=t //将其赋值给候选参数值列表
j=j+1
end for
end for
S[i]= a r g m a x x i C il(xi)/g(xi) //将C中采集函数值最大的候选场景赋值给S[i]
end for
D=D∪{(S, f(S))}
end for
return D中关键场景集合H
), ArticleFig(id=1196558426961654207, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, label=null, caption=null, figureFileSmall=null, figureFileBig=null, tableContent=
输入:观测数据集D={(x(1),y(1)),…,(x(k),y(k))},分位数γ
输出:DlDgDl={} //初始化集合Dl
j=1 //从帕累托排序等级为1开始添加到集合DI
while $ \left|D_{l}\right|+\left|D_{\text {rank }(j)}\right|<=\lfloor\gamma|D|\rfloor$ //直到将等级j的场景全部加入时,不超出Dl场景的数量
Dl=DlDrank(j) //将等级j的场景全部加入
j=j+1
end while
Dl=Dl∪贪婪超体积子集选择($ D_{\text {rank }(j),}\lfloor\gamma|D|\rfloor-\left|D_{l}\right|$) //使用贪婪超体积子集选择算法得到近似最优子集,并加入到Dl
Dg=D\Dl
return (Dl,Dg)
), ArticleFig(id=1196558427100066240, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, label=表2, caption=

场景划分

, figureFileSmall=null, figureFileBig=null, tableContent=
输入:观测数据集D={(x(1),y(1)),…,(x(k),y(k))},分位数γ
输出:DlDgDl={} //初始化集合Dl
j=1 //从帕累托排序等级为1开始添加到集合DI
while $ \left|D_{l}\right|+\left|D_{\text {rank }(j)}\right|<=\lfloor\gamma|D|\rfloor$ //直到将等级j的场景全部加入时,不超出Dl场景的数量
Dl=DlDrank(j) //将等级j的场景全部加入
j=j+1
end while
Dl=Dl∪贪婪超体积子集选择($ D_{\text {rank }(j),}\lfloor\gamma|D|\rfloor-\left|D_{l}\right|$) //使用贪婪超体积子集选择算法得到近似最优子集,并加入到Dl
Dg=D\Dl
return (Dl,Dg)
), ArticleFig(id=1196558427297198529, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, label=null, caption=null, figureFileSmall=null, figureFileBig=null, tableContent=
输入:选择集合Di,待选择数量Ns
输出:近似最优子集。
Ds={}
for all(x,y)∈Di do //计算所有属于集合Di的体积贡献度
Cy=IIH({y})
end for
for j=1, 2,,Ns do
(x′,y′)= a r g m a x ( x , y ) D i(Cy) //选取超体积指标最大的加入近似最优子集
Di=Di\{(x′,y′)}
for all(x,y)∈Di do
Cy=IIH(Ds∪{(max(y1,y1′),…,max(ym,ym′))}-IIH(Ds) //更新超体积指标
end for
Ds=Ds∪{(x′,y′)}
end for
return Ds
), ArticleFig(id=1196558427389473218, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, label=表3, caption=

贪婪超体积子集选择

, figureFileSmall=null, figureFileBig=null, tableContent=
输入:选择集合Di,待选择数量Ns
输出:近似最优子集。
Ds={}
for all(x,y)∈Di do //计算所有属于集合Di的体积贡献度
Cy=IIH({y})
end for
for j=1, 2,,Ns do
(x′,y′)= a r g m a x ( x , y ) D i(Cy) //选取超体积指标最大的加入近似最优子集
Di=Di\{(x′,y′)}
for all(x,y)∈Di do
Cy=IIH(Ds∪{(max(y1,y1′),…,max(ym,ym′))}-IIH(Ds) //更新超体积指标
end for
Ds=Ds∪{(x′,y′)}
end for
return Ds
), ArticleFig(id=1196558427448193475, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, label=null, caption=null, figureFileSmall=null, figureFileBig=null, tableContent=
场景要素 属性 分布范围
天气 云量/% [40,100]
降雨量/mm·m-3 [40,100]
积水量/% [20,40]
光照角度/(°) [-90,90]
雾浓度/% [20,35]
雾能见度/m [0,20]
交通参与者 交通密度/10辆∙km-1 [1,10]
), ArticleFig(id=1196558427527885252, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, label=表4, caption=

场景要素参数分布

, figureFileSmall=null, figureFileBig=null, tableContent=
场景要素 属性 分布范围
天气 云量/% [40,100]
降雨量/mm·m-3 [40,100]
积水量/% [20,40]
光照角度/(°) [-90,90]
雾浓度/% [20,35]
雾能见度/m [0,20]
交通参与者 交通密度/10辆∙km-1 [1,10]
), ArticleFig(id=1196558427590799813, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=EN, label=null, caption=null, figureFileSmall=null, figureFileBig=null, tableContent=
方案 搜索
场景
/个
关键
场景
/%
搜索
时间
/min
相邻场景
欧式距离
之和
关键场景
平均时间
间隔/min
Random Search 100 18 483 1 963.51 26.83
BallTree-RNS 100 27 490 1 713.37 18.15
TPE 100 56 534 848.17 9.54
MOTPE 100 37 529 1 294.10 14.30
), ArticleFig(id=1196558427657908678, tenantId=1146029695717560320, journalId=1189621681917173762, articleId=1196058110214320506, language=CN, label=表5, caption=

不同搜索方法结果对比

, figureFileSmall=null, figureFileBig=null, tableContent=
方案 搜索
场景
/个
关键
场景
/%
搜索
时间
/min
相邻场景
欧式距离
之和
关键场景
平均时间
间隔/min
Random Search 100 18 483 1 963.51 26.83
BallTree-RNS 100 27 490 1 713.37 18.15
TPE 100 56 534 848.17 9.54
MOTPE 100 37 529 1 294.10 14.30
)], attaches=null, journal=Journal(id=1149693407745847311, delFlag=0, nameCn=汽车技术, nameEn=Automobile Technology, nameHistory1=null, nameHistory2=null, issn=1000-3703, eissn=null, cn=22-1113/U, coden=null, periodic=0, language=CN, oaType=null, ccby=null, superviseOffice=null, ownerOffice=null, pubOffice=null, editorOffice=null, officeType=null, aims=null, clcCode=null, officeProv=null, officeCity=null, officeAddr=null, officeZip=null, officeEmail=null, officePhone=null, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=rYFtDx/CU9+iX8QTM0ckbw==, journalPrice=null, startedYear=null, abbrevIsoEn=null, journalRemark=null, publicationField=null, createdTime=1752037868679, updatedTime=1761735668047, createdBy=18614031015, updatedBy=13701087609, firstLetterCn=A, firstLetterEn=A, subjectCode=Engineering, subjectName=Engineering, subjectCodeEn=Engineering, subjectNameEn=null, picCn=rYFtDx/CU9+iX8QTM0ckbw==, picEn=oFT2NmUwKPUjZ27C1+d9pw==, jcr=null, cjcr=null, exts=[JournalExt(id=1190368926564450443, language=CN, name=汽车技术, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1761735668078, updatedTime=1761735668078, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://qcjs.cbpt.cnki.net/index.aspx?t=1, submissionEditorUrl=https://qcjs.cbpt.cnki.net/index.aspx?t=3, submissionReviewUrl=https://qcjs.cbpt.cnki.net/index.aspx?t=2, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1190368926618976396, language=EN, name=Automobile Technology, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1761735668091, updatedTime=1761735668091, createdBy=13701087609, updatedBy=13701087609, submissionGuidelinesUrl=, submissionAuthorUrl=https://qcjs.cbpt.cnki.net/index.aspx?t=1, submissionEditorUrl=https://qcjs.cbpt.cnki.net/index.aspx?t=3, submissionReviewUrl=https://qcjs.cbpt.cnki.net/index.aspx?t=2, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1189621681917173762, websiteList=[Website(id=1189624193747526544, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1189621681917173762, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/qcjs/CN, language=CN, createTime=1761558109939, createBy=18614031015, updateTime=1761558140534, updateBy=18614031015, name=汽车技术-中, tplId=1146099689490845704, title=汽车技术, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1189625424704451180, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193747526544, code=articleTextType, value=kx, createTime=1761558403421, updateTime=1761558403421, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625424675091049, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193747526544, code=banner, value=null, createTime=1761558403414, updateTime=1761558403414, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625424733811311, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193747526544, code=grayFlag, value=0, createTime=1761558403428, updateTime=1761558403428, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625424658313832, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193747526544, code=logo, value=https://castjournals.cast.org.cn/joweb/qcjs/CN/file/pic?fileId=7En9rzX2QCa/1J8NnKt/Fg==, createTime=1761558403410, updateTime=1761558403410, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625424746394225, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193747526544, code=minRunFlag, value=0, createTime=1761558403431, updateTime=1761558403431, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625424691868267, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193747526544, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/qcjs/CN/file/pic, createTime=1761558403418, updateTime=1761558403418, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625424742199920, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193747526544, code=silenceFlag, value=0, createTime=1761558403430, updateTime=1761558403430, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625424683479658, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193747526544, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1761558403416, updateTime=1761558403416, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625424712839789, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193747526544, code=themeColor, value=null, createTime=1761558403423, updateTime=1761558403423, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625424725422702, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193747526544, code=themeStyle, value=null, createTime=1761558403426, updateTime=1761558403426, creator=18614031015, updator=18614031015)]), Website(id=1189624193869161363, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1189621681917173762, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/qcjs/EN, language=EN, createTime=1761558109967, createBy=18614031015, updateTime=1761558340679, updateBy=18614031015, name=汽车技术-英文, tplId=1146101810881728533, title=Automobile Technology, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1189625550722311064, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193869161363, code=articleTextType, value=kx, createTime=1761558433466, updateTime=1761558433466, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625550688756629, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193869161363, code=banner, value=null, createTime=1761558433458, updateTime=1761558433458, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625550739088283, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193869161363, code=grayFlag, value=0, createTime=1761558433470, updateTime=1761558433470, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625550676173716, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193869161363, code=logo, value=https://castjournals.cast.org.cn/joweb/qcjs/EN/file/pic?fileId=7En9rzX2QCa/1J8NnKt/Fg==, createTime=1761558433455, updateTime=1761558433455, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625550751671197, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193869161363, code=minRunFlag, value=0, createTime=1761558433473, updateTime=1761558433473, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625550713922455, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193869161363, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/qcjs/EN/file/pic, createTime=1761558433464, updateTime=1761558433464, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625550743282588, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193869161363, code=silenceFlag, value=0, createTime=1761558433471, updateTime=1761558433471, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625550705533846, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193869161363, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_en_623/, createTime=1761558433462, updateTime=1761558433462, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625550726505369, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193869161363, code=themeColor, value=null, createTime=1761558433467, updateTime=1761558433467, creator=18614031015, updator=18614031015), WebsiteProps(id=1189625550734893978, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189624193869161363, code=themeStyle, value=null, createTime=1761558433469, updateTime=1761558433469, creator=18614031015, updator=18614031015)])], journalTitle=汽车技术, weixinUrl=null, journalUrl=null, iacademicId=null, status=1, seqNo=null, journalTitleEn=Automobile Technology, journalPhotoCn=rYFtDx/CU9+iX8QTM0ckbw==, journalPhotoEn=oFT2NmUwKPUjZ27C1+d9pw==, journalFirstLetter=A, journalRecommend=null, journalNew=null, journalCollection=null, jcrJf=null, cjcrJf=null, jcrJfStr=null, cjcrJfStr=null, submissionFirstDecision=null, sciSubjectClassification=null, casSubjectClassification=null, citeScore=null, totalCitationFrequency=null, icpCode=null, psCode=null, advertisingLicenseCode=null, copyrightInformation=null, country=null, option=, provinceCode=null, provinceName=null, collectFlag=false), detailUrlCn=https://castjournals.cast.org.cn/joweb/qcjs/CN/10.19620/j.cnki.1000-3703.20240092, detailUrlEn=https://castjournals.cast.org.cn/joweb/qcjs/EN/10.19620/j.cnki.1000-3703.20240092, pdfUrlCn=https://castjournals.cast.org.cn/joweb/qcjs/CN/PDF/10.19620/j.cnki.1000-3703.20240092, pdfUrlEn=https://castjournals.cast.org.cn/joweb/qcjs/EN/PDF/10.19620/j.cnki.1000-3703.20240092, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, reference=null)
收藏切换
基于树结构Parzen估计器的自动驾驶仿真测试关键场景生成方法
收藏切换
PDF下载
秦琴 1 , 杨志胜 1 , 李道鑫 1 , 沈知玮 2 , 曹晓琳 3
汽车技术 | 2025,(5): 39-46
收起
收藏切换
汽车技术 | 2025, (5): 39-46
基于树结构Parzen估计器的自动驾驶仿真测试关键场景生成方法
全屏
秦琴1, 杨志胜1 , 李道鑫1, 沈知玮2, 曹晓琳3
作者信息
  • 1 上海第二工业大学智能制造与控制工程学院,上海 201209
  • 2 新南威尔士大学电气工程与通信学院,澳大利亚 悉尼 2052
  • 3 吉林大学汽车工程学院,长春 130015

通讯作者:

杨志胜(2001—),硕士,主要研究方向为自动驾驶、汽车电子、软件测试,
Automatic Generation Method of Autonomous Driving Simulation Test Scenarios Based on Tree-Structured Parzen Estimator
Qin Qin1, Zhisheng Yang1 , Daoxin Li1, Zhiwei Shen2, Xiaolin Cao3
Affiliations
  • 1 Shanghai Polytechnic University, School of Intelligent Manufacturing and Control Engineering, Shanghai 201209
  • 2 School of Electrical Engineering and Telecommunications, University of New South Wales, Sydney 2052
  • 3 College of Automotive Engineering, Jilin University, Changchun 130015
出版时间: 2025-05-24 doi: 10.19620/j.cnki.1000-3703.20240092
文章导航
收藏切换

针对高维空间中生成关键场景场景数量呈指数增长,传统人工构造或随机搜索方法难以兼顾覆盖率与效率问题,提出基于单目标树结构Parzen估计器(TPE)和多目标树结构Parzen估计器(MOTPE)的搜索方法。通过CARLA模拟器搭建软件在环自动化仿真测试框架,以天气要素为例,对比不同搜索算法的关键场景生成效果。试验结果表明:基于TPE的搜索方法和基于MOTPE的方法相较于随机搜索方法生成的关键场景数量分别提高3.11倍和2.06倍,MOTPE方法的场景质量方面是TPE的1.53倍,配合场景自动化生成与测试框架,可有效解决场景数量爆炸问题,发现具有高测试价值的场景。

自动驾驶  /  场景生成  /  TPE  /  MOTPE  /  CARLA

In view of the exponential increase in the number of key scene scenarios generated in high-dimensional space, and the difficulty of traditional artificial construction or random search methods to balance coverage and efficiency, this paper proposes a search method based on single-objective Tree structure Parzen Estimator (TPE) and Multi-ObjectiveTree structure Parzen Estimator (MOTPE). A software-in-the-loop automated simulation testing framework is built by using the CARLA simulator. Taking weather elements as an example, the critical scenario generation effects of the different search algorithms are compared. The experimental results indicate that the TPE-based search method and the MOTPE-based method increase the number of key scenarios generated by 3.11 times and 2.06 times, respectively, compared to the random search method. The MOTPE method is 1.53 times better than TPE in terms of scenario quality. When combined with scenario automaed generation and testing frameworks, these methods effectively address the issue of exploding scenario numbers, allowing for the discovery of scenarios with high testing value.

Autonomous driving  /  Scenario generation  /  Tree-structured Parzen Estimator (TPE)  /  Multi-Objective Tree-structured Parzen Estimator (MOTPE)  /  CARLA
秦琴, 杨志胜, 李道鑫, 沈知玮, 曹晓琳. 基于树结构Parzen估计器的自动驾驶仿真测试关键场景生成方法. 汽车技术, 2025 , (5) : 39 -46 . DOI: 10.19620/j.cnki.1000-3703.20240092
Qin Qin, Zhisheng Yang, Daoxin Li, Zhiwei Shen, Xiaolin Cao. Automatic Generation Method of Autonomous Driving Simulation Test Scenarios Based on Tree-Structured Parzen Estimator[J]. Automobile Technology, 2025 , (5) : 39 -46 . DOI: 10.19620/j.cnki.1000-3703.20240092
随着车辆系统复杂性的提升,交通环境变化、驾驶任务多样性等因素成为自动驾驶车辆测试评价面临的全新挑战[1]。由于传统道路场地测试的经济与时间成本较高,且测试可重复性弱,已经无法满足极端场景的测试需求。基于场景的自动驾驶虚拟仿真测试方法在效率、测试成本等方面具有巨大优势,可成为自动驾驶车辆场景验证的重要手段[2-3]
为了克服传统场景测试方法的盲目性和低效问题,Ramakrishna等[4]基于KD树(K-dimensional Tree,Kd-Tree)结构改进随机近邻搜索算法和基于引导贝叶斯优化的关键场景生成方法[5],虽然能够显著提升关键场景的生成效率,但基于引导贝叶斯优化的关键场景生成方法的搜索时间是随机近邻搜索方法的2.7倍。文献[6]提出基于球树(Ball-Tree)算法优化随机近邻搜索(BallTree-Random Neighborhood Search,BallTree-RNS)算法,使随机近邻算法搜索时间明显降低。邢星宇等[7]提出基于高斯过程的贝叶斯优化的关键场景生成方法,和基于遗传算法的关键场景生成方法,前者生成关键场景的数量是后者的0.76倍,但关键场景的离散程度是后者的28.7倍。与传统场景测试方法不同,基于优化搜索方法生成的场景针对性更强,且在开发阶段能够进行快速迭代优化,弥补基于数据分析方法的不足。但在高维参数空间中,此类方法的计算复杂度较高。
因此,本文基于树结构Parzen估计器[8](Tree-structured Parzen Estimator,TPE)与多目标树结构Parzen估计器[9](Multi-Objective Tree-structured Parzen Estimator,MOTPE)提出关键场景优化搜索方法,使用开源仿真模拟器CARLA低成本搭建场景自动化生成与测试框架,通过对比随机搜索方法与BallTree-RNS方法,验证本文方法的有效性。
自动驾驶的行驶场景具有复杂多样、难以预测等特性,需要依据功能场景元素,结合经验知识,确定逻辑场景参数。由于场景要素众多,参数空间通常是高维连续的,且随机生成的场景风险等级较低,无法保证场景的覆盖度。若测试全部场景,虽然能够满足场景的全覆盖,但会生成大量无价值场景,导致测试效率降低,不具有可操作性。同时,受模拟器的不确定性以及待测对象的黑盒特性影响,难以依靠人工调参保证场景的覆盖率。
本文基于序列模型的优化(Sequential Model-Based Optimization,SMBO)方法,将关键场景生成问题转化为目标函数最小化的黑盒优化问题,结合自动化测试框架不断迭代优化,搜索关键价值场景,其框架如图1所示。
SMBO算法将代理模型Mt近似替代真实的目标函数f,以降低目标函数的评估成本。根据逻辑场景的参数空间D、参数间约束关系R,将搜索空间限定为图形结构。在每轮测试中,采集函数S在约束条件R下,借助代理模型对目标函数的近似评估能力,对参数空间D进行评估,并选择一个最优的候选点x*,作为本轮仿真测试的场景。重复上述过程,代理模型达到最大迭代次数后结束测试,最终生成关键场景集合H和传感器采集数据集P
由于自动驾驶仿真测试对于场景的定义未统一,本文根据文献[10]定义场景,将场景按照抽象程度划分为3类,分别为:
a. 功能场景:抽象程度最高,由自然语言描述功能性需求。例如“在雨天,车辆经过十字路口”。
b. 逻辑场景:用于描述参数空间及各参数见的逻辑关系。例如,光照角度的范围为[-90°,90°],降水量的范围为[100 mm]。
c. 具体场景:在参数空间中采样合理、具体的数值。例如,光照角度为45°、降水量为50 mm的天气场景。
利用场景要素间约束关系,利用图结构约束搜索空间,确定参数范围后进行采样,如图2所示。例如,在天气场景中,当雾浓度大于0时,可对雾能见度采样;当雾浓度等于0时,不仅增加了搜索时间,且无实际意义。
TPE基于SMBO的全局优化算法,结合预期改进(Expected Improvement,EI)准则,通过密度估计方法重构目标函数优化过程。
假设逻辑场景为X,目标函数fXRX到实数集合R的映射,在代理模型M中,当f(x)(负向)超过阈值y*时,逻辑场景参数xiXi的期望改进值为:
n E I , y * ( x i ) = - y * ( y * - y ) p ( y x i ) d y
式中:p(y|xi)为代理模型M中,输入参数xi对应目标函数y的条件概率分布。
由于p(y|xi)计算成本较高,TPE通过贝叶斯公式采用反向建模策略,利用非参数密度p(xi|y)和p(y)进行建模,其中,p(xi|y)可拆分为2个密度函数:
p ( x i y ) = l ( x i )           y y * g ( x i )         y y *
式中:lg分别为表现较好和较差的场景密度分布。
l(xi)由观测值集合的子集{xi}定义,该子集满足 $\left\{x_{i}^{(j)} \in X_{i} \mid y^{(j)}\left(=f\left(x^{(j)}\right)\right)<y^{*}\right\}$,使用分位数γ确定y*,即y*满足p(y<y*)=γγ∈(0,1),根据经验本文选择γ=0.15。
假设Dx={xi|(x,y)∈Dl}或Dx={xi|(x,y)∈Dg},使用Parzen估计器构建l(xi)或g(xi)的密度为:
p ( x i ) = x i ' D x w x i ' k ( x i , x i ' ) + w p k ( x i , x p ) x i ' D x w x i ' + w p
式中: w x i '为具体场景的权重,wp为固定先验xp的权重, k ( x i , x i ' )为样本点 x i '处放置的截断高斯核函数。
使用截断高斯核函数构建密度估计器Ntrunc(μ,σ,a,b),其中,σ为带宽,μ为均值,ab分别为定义域的下界和上界。通过上述建模方式,得到参数xi的期望改进值:
$n_{\mathrm{EI}, y^{*}}\left(x_{i}\right)=\int_{-\infty}^{y^{*}}\left(y^{*}-y\right) p\left(y \mid x_{i}\right) \mathrm{d} y \propto\left(\gamma+\frac{g\left(x_{i}\right)}{l\left(x_{i}\right)}(1-\gamma)\right)^{-1}$
因此,基于TPE的搜索算法伪代码如表1所示。当某个采样场景重复出现,极易导致测试效率降低。为了保证测试场景多样性与算法稳定性,本文基于球树算法,综合考虑场景的关键程度以及其在指定邻域内的样本密度,在保持测试多样性的同时,提升算法的稳定性和效率:
w D l x ' = f ( x ' ) × ( L - B a l l T r e e K n n ( x ' , d ) ) w D g x ' = f ( x ' ) × B a l l T r e e K n n ( x ' , d ) L = m a x ( B a l l T r e e K n n ( D , d ) ) + C
式中:f(x')为具体场景x'的评价函数,BallTreeKnn(x)为范围d中场景x的数量,L为所有场景的K近邻最大值与常数C之和。
式(5)中,常数C越小,偏向于搜索多样性的场景,反之,则偏向于选择更危险的场景。
MOTPE方法设置观测值集合为D={(x(1),y(1)),…,(x(k),y(k))},对于每个逻辑场景参数xiXi,则p(xi|y)为:
p ( x i y ) = l ( x i )           ( y Y * ) ( y | | Y * ) g ( x i )         y Y *
Y*满足p((yY*)∪(y||Y*))=γ的场景,本文根据经验选取γ=0.15,yY*表示在非支配排序中,所有场景Y*y支配,y||Y*表示yY*不存在支配关系,如图3所示。
鉴于目标函数为多目标,MOTPE在观测集划分策略方面不同于TPE方法。TPE的样本划分基于单一目标函数值的排序,而MOTPE则需要依据多目标优化中的帕累托支配关系进行处理,即MOTPE对观测集进行非支配排序,将非支配等级较高的样本集合优先加入Dl,再引入贪婪超体积子集选择(Hypervolume Subset Selection Problem,HSSP)策略,选择近似最优子集加入Dl,剩余样本划分为Dg。场景划分伪代码如表2所示。
使用贪婪HSSP算法获得近似最优子集,保证所选子集的超体积值不低于最优超体积的(1-e-1)倍[11],算法伪代码如表3所示。
本文使用的权重策略与TPE相同,由于传统期望改进策略不适用于多目标优化,所以将期望超体积改进(Expected Hypervolume Improvement,EHVI)作为采集函数,得到最终期望超体积改进:
$\begin{aligned} n_{\mathrm{EHVI}, Y^{*}}\left(x_{i}\right)= & \int\left(n_{\mathrm{IH}}\left(Y^{*} \cup\{y\}\right)-n_{\mathrm{IH}}\left(Y^{*}\right)\right) p\left(y \mid x_{i}\right) \mathrm{d} y \propto \\ & \left(\gamma+\frac{g\left(x_{i}\right)}{l\left(x_{i}\right)}(1-\gamma)\right)^{-1} \end{aligned}$
式中: n I H为超体积指标函数。
本文基于场景的测试方法,利用开源仿真模拟器CARLA、CARLA Leaderboard模块(CARLA挑战赛基准评测模块)构建软件在环测试框架。该框架可保存危险场景数据,用于模型的二次训练,同时,在系统开发的初级阶段在一定程度上消除模型隐患。
测试框架包括场景生成模块、评价模块和仿真模块,如图4所示。测试开始时,由场景生成模块生成具体的可执行场景文件,再经过仿真模块加载场景文件以及模型文件,并开始运行仿真测试,本轮测试结束后,通过评价模块获得测试结果,并返回场景生成模块,继续新一轮的场景生成与测试。
CARLA支持多种传感器(如激光雷达、摄像头、GPS等)配置,具备模块化应用程序编程接口。便于场景与测试流程的集成[12]。通过结合Leaderboard模块,使用CARLA Scenario Runner提供的仿真世界的信息,评估被测对象在模拟器中的运行状态,实现自动驾驶的性能评估,构建软件在环的自动化仿真测试环境,仿真场景如图5所示。
鉴于测试场景多为高纬度场景,各场景要素间为有机关联的整体。因此,在场景生成模块中,使用YAML语言设计采样约束文件及场景描述文件,利用结构化描述定义逻辑场景[4]。在场景描述文件中手动配置参数及其范围,在采样约束文件中设置约束条件,从而确保有效测试场景,降低参数空间,加快搜索速度。同时,该模块集成采样器组件,在逻辑场景的参数空间中搜索具体场景。对于TPE及MOTPE算法,在约束条件下搜索高测试价值场景。
仿真开始后,采样器读取场景描述文件和采样约束文件,将逻辑场景及约束条件作为采样器的输入,采样器获得具体场景后,由场景要素解释器接收结果,合成新的可执行场景文件,进行下一轮仿真测试。
由于天气条件易影响视觉传感器的性能,当光线过强时,光线的能量无法迅速散射,导致反射光经多次折射形成眩光,进而降低图像的对比度;当光线到达成像平面时,会形成虚像点或明亮光斑,影响成像质量[13]。因此,本文以天气场景为例,基于开源高保真模拟器搭建自动化测试平台,并验证提出方法的有效性。
本文被测对象为LBC控制器模型,该模型将基于视觉的自动驾驶系统分为2个关键阶段:第一阶段利用模拟器提供的环境信息,构建一个访问环境真实状态信息(如地图布局与动态参与者位置)的专家控制器,用于指导车辆行为;第二阶段以该专家代理作为教师,训练一个仅依赖于前视图像输入的端到端驾驶策略。通过横向、纵向比例-积分-微分(Proportion Integration Differentiation,PID)控制器实现对特定路径点的跟踪,在CARLA基准测试中展示出优异的鲁棒性[14]
为了充分验证模型在不同天气场景下的表现,本文将CARLA提供的5号小镇作为测试场地,该地图逼真度高、细节丰富,设有丰富的交通标志和车道线,每轮测试选择不同路线,保证测试的复杂性,如图6所示。
试验环境选择天气和交通参与者要素,其中,天气要素包含云量、降雨量、积水量、光照角度、雾浓度、雾能见度;交通参与者包括行人与车辆,使用CARLA内部的人工智能(Artificial Intelligence,AI)行人与自动驾驶模块,并引入交通密度参数。
逻辑场景的场景要素及参数分布如表4所示,通过添加约束条件,当降雨量为[100 mm]时,光照角度为[-20°,90°],避免极端场景下无法正常执行预期的自动驾驶任务。
为了评估自动驾驶车辆的安全性,根据文献[15]的安全性评价指标,选取CARLA Leaderboard中受天气要素影响较大的评价指标构建目标函数,用以表征各场景参数组合的关键程度。将碰撞率Rc、闯红灯频率Rr、平均道路偏离距离dmean,o作为评价指标,其中,Rc反映被测对象对交通参与者的识别性能,Rr反映被测对象对交通标志的识别性能,dmean,o反映被测对象的车道线识别能力以及车道保持性能。相关公式为:
R c = T p + T v S d R r = T r S d d m e a n , o = S o S d
式中:Tp为自车与行人的碰撞次数,Tv为自车与车辆、行人的碰撞次数,Sd为行驶轨迹的距离,Tr为闯红灯次数,So为车辆偏离道路的行驶距离。
基于上述指标,本文的目标函数为:
$f=\left\{\begin{array}{l} \text { None,使用TPE算法或MOTPE算法的无效场景 } \\ -\left(R_{\mathrm{c}}+R_{\mathrm{r}}+d_{\text {mean, } o}\right) \text {, 使用TPE算法的有效场景 } \\ -R_{\mathrm{c}},-R_{\mathrm{r}},-d_{\text {mean, } o} \text {, 使用MOTPE算法的有效场景 } \end{array}\right.$
针对无效场景,车辆无法完成驾驶任务,为了保证代理模型的分布,目标函数值设置为无效值;在有效场景中,为了客观分析TPE方法和MOTPE方法的性能表现,对不同评价指标使用相同的权重,根据实际情况通过层次分析法分配不同权重,并选择集成模型进行多指标综合性评价,表征场景的关键度,将最大值优化问题转化为最小值优化问题。
本文针对逻辑场景中不同天气要素,使用随机搜索(Random Search)、球树随机邻域搜索(BallTree-Random Neighborhood Search,BallTree-RNS)方法[6]分别搜索100个仿真测试场景;TPE方法和MOTPE方法使用随机搜索方法构建初始代理模型,最大迭代100次为终止条件,各方法的搜索质量与效率如表5所示。
表5可知,Random、BallTree-RNS的搜索结果中,关键场景占比分别为18%和27%,而TPE和MOTPE搜索到更多的关键场景,占比分别为56%和37%。各方法搜索关键场景的平均间隔时间为TPE∶MOTPE∶BallTree-RNS∶Random Search =1∶1.5∶1.9∶2.8。结合不同方法统计评价指标结果,如图7所示,TPE方法生成关键危险场景的效率高于MOTPE方法,同时,二者均优于使用近邻搜索策略的BallTree-RNS和未使用优化策略的Random方法。
在场景质量方面,本文以相邻具体场景的欧式距离之和表征搜索方法对场景空间的探索能力[7,16],对于随机搜索方法,相邻具体场景欧式距离之和最高达到1 963.51,而随机近邻搜索方法在其邻域内搜索关键场景时,其分散程度降低至1 713.37。可见,随机近邻搜索方法增强了局部搜索能力的同时,也相应地牺牲了对全局场景空间的探索能力。
相较于随机搜索方法,TPE方法和MOTPE方法均采用基于代理模型和采集函数的优化策略,在搜索过程中更注重利用与探索的平衡性,通过引导采样过程集中于潜在高价值区域,故两种方法的相邻场景欧式距离之和均低于随机搜索方法和近邻搜索方法。而MOTPE方法相邻场景欧式距离为1 294.10,约为TPE方法的1.53倍。
为了分析不同搜索方法生成场景在高维参数空间的分布特性,采用主成分分析法(PCA)对场景参数向量进行降维,并提取前3个主成分进行三维可视化,空间分布如图8所示,其中,关键场景表示Leaderboard统计为失败的场景,即在该场景下车辆存在目标函数定义的违规行为,本文设置(Rc+Ro+Rr)≥0.015。
图8可知,在生成相同数量场景时,以随机搜索方法为基线,TPE和MOTPE搜索关键场景数量分别为基准方法的3.11倍和2.06倍;对于生成关键场景的平均时间间隔,随机搜索方法为26.83 min/个,TPE和MOTPE约为基线的1/3、1/2;在搜索质量方面,MOTPE方法考虑了多目标函数的实际情况,MOTPE是TPE的1.53倍,但搜索效率方较低。TPE方法和MOTPE方法的场景参数分布相较为集中,有助于测试人员分析驾驶违规的原因,而MOTPE方法表现出较强的场景覆盖能力与空间探索能力。
根据TPE与MOTPE方法在关键场景生成效果方面的差异,虽然两种方法均采用基于代理模型与采集函数的优化策略,但设计目标略有不同。TPE方法面向单目标优化,能在有限的搜索次数内快速聚焦于性能较差的样本区域,从而在固定数量的采样中获得较高比例的关键场景,表现出更高的搜索效率。MOTPE方法在优化过程中考虑多个性能指标,通过期望超体积改进策略,引导搜索过程更全面地逼近帕累托前沿,增强生成场景的多样性。但由于该方法目标函数更加分散,导致生成关键场景的比例相对较低,因而搜索效率略低于TPE方法。
本文基于树结构Parzen估计器提出关键场景生成方法,缓解了高维空间样本难以开发的问题,有效避免了传统场景测试方法的随机性和低效率。未来,可引入离散型参数,构建更符合真实交通环境的多维复杂场景参数空间,以提高本文方法的泛化性。
参考文献 引证文献
排序方式:
[1]
孙乐乐, 黄松, 郑长友, 等. 自动驾驶仿真测试场景生成技术研究进展[J]. 计算机工程与应用, 2025, 61(1): 59-79.
SUN L L, HUANG S, ZHENG C Y, et al. Research Progress on Autonomous Driving Simulation Test Scenario Generation Technology[J]. Computer Engineering and Applications, 2025, 61(1): 59-79.
[2]
蒋拯民, 党少博, 李慧云, 等. 自动驾驶汽车场景测试研究进展综述[J]. 汽车技术, 2022(8): 10-22.
JIANG Z M, DANG S B, LI H Y, et al. A Survey on the Research Progress of Scenario-Based Testing for Autonomous Vehicles[J]. Automobile Technology, 2022(8): 10-22.
[3]
邓伟文, 李江坤, 任秉韬, 等. 面向自动驾驶的仿真场景自动生成方法综述[J]. 中国公路学报, 2022, 35(1): 316-333.
DENG W W, LI J K, REN B T, et al. A Survey on Automatic Simulation Scenario Generation Methods for Autonomous Driving[J]. China Journal of Highway and Transport, 2022, 35(1): 316-333.
[4]
RAMAKRISHNA S, LUO B T, KUHN C B, et al. ANTI-CARLA:An Adversarial Testing Framework for Autonomous Vehicles in CARLA[C]// 2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC). Macau, China: IEEE, 2022: 2620-2627.
[5]
RAMAKRISHNA S, LUO B T, BARVE Y, et al. Risk-Aware Scene Sampling for Dynamic Assurance of Autonomous Systems[C]// 2022 IEEE International Conference on Assured Autonomy (ICAA). Fajardo, PR, USA: IEEE, 2022: 107-116.
[6]
秦琴, 谷文军. Ball Tree优化的自动驾驶仿真测试场景生成方法[J]. 计算机应用研究, 2023, 40(9): 2781-2784+2791.
QIN Q, GU W J. Ball Tree Optimized Automatic Driving Simulation Test Scenario Generation Method. Application Research of Computers, 2023, 40(9): 2781-2784+2791.
[7]
邢星宇, 吴旭阳, 刘力豪, 等. 基于目标优化的自动驾驶决策规划系统自动化测试方法[J]. 同济大学学报(自然科学版), 2021, 49(8): 1162-1169.
XING X Y, WU X Y, LIU L H, et al. Automatic Testing Method Based on Optimization Algorithms for the Decision and Planning System of Autonomous Vehicles[J]. Journal of Tongji University (Natural Science), 2021, 49(8): 1162-1169.
[8]
BERGSTRA J, BARDENET R, BENGIO Y, et al. Algorithms for Hyper-Parameter Optimization[C]// Proceedings of the 25th International Conference on Neural Information Processing Systems. Granada Spain:PMLR, 2011: 2546-2554.
[9]
OZAKI Y, TANIGAKI Y, WATANABE S, et al. Multiobjective Tree-Structured Parzen Estimator[J]. Journal of Artificial Intelligence Research, 2022(73): 1209-1250.
[10]
MENZEL T, BAGSCHIK G, MAURER M. Scenarios for Development,Test and Validation of Automated Vehicles[C]// 2018 IEEE Intelligent Vehicles Symposium (IV). Changshu, China: IEEE, 2018: 1821-1827.
[11]
GUERREIRO A P, FONSECA C M, PAQUETE L. Greedy Hypervolume Subset Selection in Low Dimensions[J]. Evolutionary Computation, 2016, 24(3): 521-544.
[12]
PIRRI P, PAHL C, IOINI N E, et al. Towards Cooperative Maneuvering Simulation:Tools and Architecture[C]// 2021 IEEE 18th Annual Consumer Communications & Networking Conference (CCNC). Las Vegas, NV, USA: IEEE, 2021.
[13]
白雪松, 邓伟文, 任秉韬, 等. 一种自动驾驶仿真场景要素的提取方法[J]. 汽车工程, 2021, 43(7): 1030-1036+1065.
BAI X S, DENG W W, REN B T, et al. An Extraction Method of Scenario Elements for Autonomous Driving Simulation[J]. Automotive Engineering, 2021, 43(7): 1030-1036+1065.
[14]
CHEN D, ZHOU B, KOLTUN V, et al. Learning by Cheating[C]// Proceedings of the Conference on Robot Learning. Auckland, New Zealand: PMLR, 2020(100): 66-75.
[15]
XU C J, DING W H, LV W J, et al. Safebench: A Benchmarking Platform for Safety Evaluation of Autonomous Vehicles[C]// 26th Conference on Neural Information Processing Systems Datasets and Benchmarks Track. New Orleans, USA: PMLR, 2022(35): 25667-25682.
[16]
FENG S, FENG Y H, YU C H, et al. Testing Scenario Library Generation for Connected and Automated Vehicles, Part I: Methodology[J]. IEEE Transactions on Intelligent Transportation Systems, 2020, 22(3): 1573-1582.
2025年第卷第5期
PDF下载
268
117
引用本文
BibTeX
文章信息
doi: 10.19620/j.cnki.1000-3703.20240092
  • 首发时间:2025-11-14
  • 出版时间:2025-05-24
补充材料
相关文章
文章信息
作者
出版历史
  • 修回日期:2024-04-15
基金
作者信息
    1 上海第二工业大学智能制造与控制工程学院,上海 201209
    2 新南威尔士大学电气工程与通信学院,澳大利亚 悉尼 2052
    3 吉林大学汽车工程学院,长春 130015

通讯作者:

杨志胜(2001—),硕士,主要研究方向为自动驾驶、汽车电子、软件测试,
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/qcjs/CN/10.19620/j.cnki.1000-3703.20240092
分享至
全文二维码

扫描看全文

引用本文
BibTeX
本文的引用情况
2种不同金属材料的力学参数

Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)

Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏