典型城市道路人车碰撞危险场景提取  

Extraction of Pedestrian and Vehicle Collision Scenes on Typical Urban Roads

作　　者：邹晋彬 倪兴泷 但舒灵 ZOU Jin-Bin;NI Xing-long;DAN Shu-ling(Xihua University,Chengdu,Sichuan 610039,China)

机构地区：[1]西华大学,四川成都610039

出　　处：《黑龙江交通科技》2022年第3期165-166,共2页Communications Science and Technology Heilongjiang

基　　金：国家级大学生创新创业训练项目(S202010650001);西华大学西华杯项目(2020143)。

摘　　要：为了减少自动驾驶汽车在城市道路测试的开发成本和开发周期,并为城市道路交通规划提供参考,基于国家车辆事故深度调查体系(NAIS)数据库,筛选出了144起城市道路汽车与行人碰撞的工况样本,对筛选出的交通事故案例聚类分析,通过卡方检验提取危险场景的典型特征,最终得到4类符合中国城市道路汽车与行人碰撞的典型危险场景库,为构建城市智能交通方案提供参考。In order to reduce the development cost and development cycle of autonomous vehicle testing on urban roads,and provide a reference for urban road traffic planning,based on the National Vehicle Accident In-depth Investigation System(NAIS)database,144 urban road vehicle and pedestrian collisions were screened out based on the cluster analysis of the selected traffic accident cases,the typical characteristics of the dangerous scenes are extracted through the chi-square test,and finally four types of typical dangerous scene libraries in line with the collision of vehicles and pedestrians on Chinese urban roads are obtained,in order to provides reference for construct urban intelligent transportation.

关 键 词：自动驾驶 人车碰撞 危险场景 聚类分析 

分 类 号：U492[交通运输工程—交通运输规划与管理]