基于典型非机动车事故的自动驾驶汽车测试场景构建方法研究  

作　　者：严慈磊 应朝阳[1] 孙巍[1] 范志翔 

机构地区：[1]公安部交通管理科学研究所

出　　处：《道路交通科学技术》2022年第1期15-22,共8页Road Traffic Science & Technology

基　　金：中央公益科研所项目“国家智能交通综合测试基地智能车路系统(车联网)测试关键技术研究”(项目编号:2020SJA02)资助。

摘　　要：基于交通事故卷宗、交通事故视频信息数据,研究机非混行交通环境下典型交通事故形态,构建了面向机非混行交通环境下的自动驾驶汽车测试场景,旨在针对我国较为特殊的机非混行环境下的自动驾驶汽车的测试场景及测试评价方法提供参考。本文首先分析了自动驾驶测试场景的构建需求,建立交通事故数据筛选标准,得到133例可用于构建自动驾驶汽车测试场景的机动车与非机动车交通事故数据集;其次基于《中华人民共和国道路交通安全法》行驶要求,对133例交通事故的发生地点、车辆行为、道路类型、环境光线等方面进行解构分析;最后通过聚类分析,建立了5类典型的自动驾驶测试场景模型,并分析了不同场景模型的关键要素,为实际道路测试提供理论指导。Based on traffic accident dossiers and traffic accident video information data,the paper studies the typical traffic accident patterns in the mixed traffic(the motor vehicle mixed with the non-motor vehicle)environment and constructs a test scenario for autonomous vehicles in the mixed traffic environment with the aim of providing reference for test scenarios and test evaluation methods of autonomous vehicles in the mixed traffic environment.Firstly,the paper analyzes the requirements for the construction of autonomous driving test scenarios,establishes traffic accident data screening criteria and obtains data sets of 133 motor vehicle traffic accident cases and non-motor vehicle traffic accidents that can be used to construct autonomous vehicle test scenarios.Then,the paper deconstructs and analyzes accident elements of 133 cases such as the location of accidents,vehicle behaviors,road types,ambient light and so on in accordance with the driving requirements of the Road Traffic Safety Laws of the People's Republic of China.Finally,through the cluster analysis,five typical autopilot test scene models are established and key elements of different scene models are analyzed,which provides theoretical guidance for actual road testing.

关 键 词：机非混行 数据集 解构分析 聚类分析 测试场景模型 

分 类 号：U46[机械工程—车辆工程]