面向自动驾驶的典型汽车与行人事故冲突场景研究  

Trace analysis for the typical precrash scenario between car vehicle and pedestrian caused by the automatic driving

作　　者：谭正平[1,2,3] 车瑶栎 肖凌云 李平飞 张巧玉[1] 徐进 TAN Zheng-ping;CHE Yao-yue;XIAO Ling-yun;LI Ping-fei;ZHANG Qiao-yu;XU Jin(National Experiment Teaching Demonstration Center of Automotive Engineering,Xihua University,Chengdu 610039,China;College of Traffic&Transportation,Chongqing Jiaotong University,Chongqing 400074,China;Sichuan Xihua Jiaotong Forensic Science Center,Chengdu 610039,China;State Administration for Market Regulation Defective Product Administrative Center(DPAC),Beijing 100101,China)

机构地区：[1]西华大学汽车工程国家级实验教学示范中心,成都610039 [2]重庆交通大学交通运输学院,重庆400074 [3]四川西华交通司法鉴定中心,成都610039 [4]国家市场监督管理总局缺陷产品管理中心,北京100101

出　　处：《安全与环境学报》2021年第4期1573-1582,共10页Journal of Safety and Environment

基　　金：国家车辆事故深度调查体系项目(192273);上海汽车工业科技发展基金项目(1828);西华大学汽车与交通工程实验教学团队/汽车服务工程教学团队建设项目。

摘　　要：为了从交通事故数据中提取典型汽车与行人冲突场景,应用于自动驾驶汽车研发测试,首先将行人事故冲突场景的组成变量按主车、行人、道路、环境4个类别进行划分,利用群决策理论分析了各类别下属变量的重要性,提取重要变量并根据其取值定义了汽车与行人冲突的基础场景。结合国家车辆事故深度调查体系(NAIS)中行人事故数据,统计得到46类基础场景,分析了中国行人冲突基础场景的特征,并提取了7种典型的基础场景,就其中2类最典型场景进行聚类分析,对后5类典型场景进行统计分析,获得了16类典型汽车与行人冲突场景的详细描述。研究结果可为系统全面地研究我国自动驾驶汽车行人自动紧急制动(AEB)测试场景提供参考。The present paper is devoted to a trace and investigative study so as to obtain the typical pre-crash scenarios of the vehicle-to-pedestrian traffic accidents by using 441 accident cases. Needless to say,in the said cases,the pedestrians may have suffered serious injuries or even sacrificed their lives due to the collisions,as recorded databases by the National Automobile Accident In-depth Investigation System( NAIS). To illustrate and describe such unhappy accidents in a more or less clear-cut way,first of all,it would be convenient to divide the following12 pedestrian accident scenario variables into 4 categories: the host vehicle,the pedestrian,the road condition,and the on-spot environmental condition,on which the respective responsibilities of the variables concerned can be respectively discriminated via the group decision theory. And,in conclusion,the maneuvers,the transportation movement,the type of road and the line of sight can serve as the key factor variables respectively. And,furthermore,the first 3 factors combination can be directly chosen as related to the scenario constitutions defined by the basic scenarios of all the accidents between the vehicles and the pedestrians in the NAIS database with the entire constitution of 46 basic scenarios. Among them,the top 2 and top 7 basic scenarios have statistically arrived at 68. 9% and 84. 6% accordingly.And,further,we have chosen 4 variables: the pedestrian speed,the lighting conditions,the line of sight and the weather and climatic conditions for the subsequent analysis so as to take away more detailed descriptions of typical scenarios. As a result,the correlated data of the top 2 basic scenarios have been analyzed in a clustered way. The data of the third basic scenario to the seventh one can be analyzed via the direct statistics. And,so,16 typical car vehicles have been described in detail in the relation to the pedestrian pre-crash ones. And,in conclusion,this paper has analyzed the data distribution features of 46 basic scenarios,showing the differen

关 键 词：安全系统学 车辆事故 行人 测试场景 自动驾驶 自动紧急制动(AEB) 聚类分析 

分 类 号：X913[环境科学与工程—安全科学]