城市环形交叉口智能汽车接管过程事故风险预测  

Accident risk prediction of intelligent vehicle takeover process at urban roundabout

作　　者：刘擎超[1,2] 徐天宇 熊晓夏 赵晶娅 蔡英凤 LIU Qingchao;XU Tianyu;XIONG Xiaoxia;ZHAO Jingya;CAI Yingfeng(Automotive Engineering Research Institute,Jiangsu University,Zhenjiang Jiangsu 212013,China;School of Mechanical and Aerospace Engineering,Nanyang Technological University,Singapore 639798,Singapore;School of Automotive and Traffic Engineering,Jiangsu University,Zhenjiang Jiangsu 212013,China;School of Transportation,Southeast University,Nanjing Jiangsu 210096,China)

机构地区：[1]江苏大学汽车工程研究院,江苏镇江212013 [2]南洋理工大学机械与航空航天工程学院,新加坡639798 [3]江苏大学汽车与交通工程学院,江苏镇江212013 [4]东南大学交通学院,江苏南京210096

出　　处：《中国安全科学学报》2022年第12期150-157,共8页China Safety Science Journal

基　　金：国家自然科学基金资助(51905223,U20A20331,U20A20333,52225212);江苏省重点研发计划项目(BE2020083-3,BE2019010-2,BE2021011-3);中国博士后科学基金资助(2021M690069);江苏省研究生实践创新计划项目(SJCX20_1425)。

摘　　要：为降低城市环形交叉口事故发生率,针对交叉口智能汽车接管过程中的事故风险问题,提出智能汽车事故风险预测模型。基于城市交通仿真软件(ToT),构建城市环形交叉口场景,分析智能驾驶汽车接管过程的事故数据,揭示道路区域及接管时间对智能汽车接管过程事故风险的影响机制;使用CatBoost对事故风险进行建模,以敏感性分析、曲线下面积(AUC)为评价指标,对比CatBoost与线性回归、XGBoost模型的预测性能。结果表明:速度对事故的影响重要度占比在47%以上,交叉口入口道事故率最高,且入口左侧车道事故率较右侧车道平均高8.63%左右;ToT时间在环形交叉口对汽车事故率影响约为8.5%,且环道区域的道路曲率、半径因素对事故影响因素占比小于5%;环道路段几乎不影响自动驾驶汽车在车辆接管过程的碰撞,CatBoost模型预测精度高于线性回归及XGBoost。Aiming at the accident risk in the process of intelligent vehicle takeover at the urban roundabout,a prediction model of intelligent vehicle accident risk at the roundabout was proposed to reduce the accident rate.Based on the urban traffic simulation software simulation of urban mobility(SUMO),the scene of urban roundabout was established,and the accident data of intelligent driving vehicle takeover process was analyzed to reveal the impact mechanism of road area and takeover time(ToT)on the accident risk of intelligent vehicle takeover process.CatBoost was used to model the accident risk,and the sensitivity analysis and Area Under the Curve(AUC)were used as evaluation indicators to compare the prediction performance with Linear Regression and XGBoost models.The results show that the impact of speed on the accident is more than 47%.The accident rate in the entrance lane of the intersection is the highest.The accident rate of the left lane of the entrance is 8.63%higher than that of the right lane on average.The influence of(ToT)on car accident rate at roundabouts is about 8.5%,and the proportion of road curvature and radius factors in the loop area on accident factors is less than 5%.The roundabout segment hardly affects the autonomous vehicle collision during the vehicle takeover process.The prediction accuracy of the CatBoost model is higher than that of linear regression and XGBoost.

关 键 词：环形交叉口 智能汽车 接管时间 事故分析 CatBoost 

分 类 号：X928[环境科学与工程—安全科学]