十字路口视觉障碍下车辆-VRUs碰撞风险量化评估方法

doi:10.19562/j.chinasae.qcgc.2025.09.008

摘要/Abstract

摘要：

针对十字路口视觉障碍场景下车辆与弱势道路使用者（VRUs）的碰撞风险，本文提出了一种融合道路环境特征的行车风险评估方法。基于VRU-TRAVi（vulnerable road users traffic accident database with video）数据库中的831例事故视频，通过K-modes聚类提取3类典型场景：通行信号灯、无信号灯及警示信号灯交叉口。通过差异性分析，揭示了车辆与障碍物的速度（VSpd、OSpd）、加速度（VAcc、OAcc）与道路环境特征的显著性关联。基于聚类场景中运动学参数的中位数设定安全阈值，并结合道路特征权重构建了风险评估模型（U_rfr）。结果表明：在通行信号灯交叉口，当障碍物速度OSpd=0，车辆速度VSpd≥45 km?h^-1、加速度VAcc≥0时行车风险最高。在无信号灯交叉口，当障碍物OSpd=0、车辆VSpd≥35 km?h^-1、VAcc≥0时行车风险最高。在警示信号灯交叉口，当障碍物速度OSpd≤10.29 km?h^-1、加速度OAcc≤0、车辆VSpd≥38 km?h^-1、VAcc≥3.74 m?s^-2时行车风险最高。模型量化了道路环境特征对运动学参数的差异化影响，可为自动驾驶车辆在复杂视觉障碍场景下的风险预测与主动控制提供理论支持。

关键词: 自动驾驶安全, 道路特征风险评估, 聚类分析, VRUs, 风险量化

Abstract:

For the collision risk between vehicles and vulnerable road users （VRUs） at intersections with visual occlusions， in this study a driving risk assessment method integrating road environment characteristics is proposed. Based on 831 accident videos from the VRU-TRAVi （Vulnerable Road Users Traffic Accident database with Video）， K-modes clustering is used to extract three typical scenarios： signalized intersections， unsignalized intersections， and warning signal intersections. Through variability analysis， the study reveals significant correlation between kinematic parameters （vehicle speed VSpd， obstacle speed OSpd， vehicle acceleration VAcc， and obstacle acceleration OAcc） and road environment features. A risk assessment model （U_rfr） is developed by setting safety thresholds based on the median values of kinematic parameters in clustered scenarios and incorporating road feature weights. The results show that： at traffic signalized intersections， the highest risk of driving occurs when the obstacle speed OSpd = 0， vehicle speed VSpd ≥ 45 km?h^-1， acceleration VAcc ≥ 0. At unsignalized intersections， the highest risk of driving occurs when the obstacle OSpd = 0， vehicle VSpd ≥ 35 km?h^-1， VAcc ≥ 0. At warning signalized intersections， the driving risk is highest when the obstacle speed OSpd ≤ 10.29 km?h^-1， acceleration OAcc ≤ 0， and the vehicle VSpd ≥ 38 km?h^-1， VAcc ≥ 3.74 m?s^-2. The model quantifies the impact of road environment features on kinematic parameters， providing a theoretical foundation for risk prediction and active control of autonomous vehicles in visually occluded intersection scenarios.

Key words: autonomous driving safety, road characterization risk assessment model, cluster analysis, VRUs, risk quantification

韩勇,张佳乐,潘迪,吴贺,徐莉. 十字路口视觉障碍下车辆-VRUs碰撞风险量化评估方法[J]. 汽车工程, 2025, 47(9): 1721-1730.

Yong Han,Jiale Zhang,Di Pan,He Wu,Li Xu. Quantitative Assessment of Vehicle-VRUs Collision Risk at Intersection with Visual Obstacle[J]. Automotive Engineering, 2025, 47(9): 1721-1730.

图/表 19

表1

表2

图1

图2

图3

表3

图4

图5

图6

表4

表5

表6

表7

表8

表9

图7

表10

表11

图8

参考文献 34

[1]	Global status report on road safety 2023［R］. Switzerland： World Health Organization， 2023： 4-9.
[2]	ZHANG L， ZHANG Z， WANG Z， et al. Chassis coordinated control for full X-by-wire vehicles-a review［J］. Chinese Journal of Mechanical Engineering， 2021， 34： 1-25.
[3]	WOLFF V A， XHOXHI E. Mitigating vulnerable road users occlusion risk via collective perception： an empirical analysis［J］. arXiv preprint arXiv：， 2024.
[4]	PEI Y， HOU L. Safety assessment and risk management of urban arterial traffic flow based on artificial driving and intelligent network connection： an overview［J］. Archives of Computational Methods in Engineering， 2024： 1-19.
[5]	HARRINGTON S， NAGARAJAN S R， LAU J. An evaluation of the sensitivity of the user-selected forward collision warning distance in a 2017 Honda CR-V［C］. SAE Paper 2023-01-0622.
[6]	韩勇，孟昕，潘迪，等. 视觉障碍下十字路口电动两轮车事故特征和典型场景分析［J］. 汽车安全与节能学报， 2023， 14（6）： 664-670.
	HAN Y， MENG X， PAN D， et al. Characteristics and typical scenarios of electric two-wheeler accidents at intersections with visual obstacles［J］. Journal of Automotive Safety and Energy， 2023，14（6）： 664-670.
[7]	TYAGI I. Threat assessment for avoiding collsions with perpendicular vehicles at intersections［C］.2021 IEEE International Conference on Electro Information Technology （EIT）. IEEE， 2021： 184-187.
[8]	XIONG X， ZHANG S， CHEN Y. Review of intelligent vehicle driving risk assessment in multi-vehicle interaction scenarios［J］. World Electric Vehicle Journal， 2023， 14（12）： 348.
[9]	WANG W， WANG W， WAN J， et al. Potential field based path planning with predictive tracking control for autonomous vehicles［C］.2019 5th International Conference on Transportation Information and Safety （ICTIS）. IEEE， 2019： 746-751.
[10]	WANG J， WU J， ZHENG X， et al. Driving safety field theory modeling and its application in pre-collision warning system［J］. Transportation Research Part C： Emerging Technologies， 2016， 72： 306-324.
[11]	MA Y， ZHANG P， HU B. Active lane-changing model of vehicle in B-type weaving region based on potential energy field theory［J］. Physica A： Statistical Mechanics and its Applications， 2019， 535： 122291.
[12]	何仁，赵晓聪，王建强. 人-车-路交互下的驾驶人风险响应度建模［J］. 中国公路学报， 2020， 33（9）： 236-250.
	HE R， ZHAO X， WANG J. Driver risk response modeling under the human-vehicle-road interaction［J］. China Journal of Highway and Transport， 2020， 33（9）： 236-250.
[13]	田野，裴华鑫，晏松，等. 车路协同环境下行车风险场模型的扩展与应用［J］. 清华大学学报（自然科学版）， 2022， 62（3）： 447-457.
	TIAN Y， PEI H， YAN S， et al. Extension and application of the vehicle-road collaborative environment driving risk field model［J］. Journal of Tsinghua University （Science and Technology）， 2022，62（3）： 447-457.
[14]	WANG M， ZHANG L， ZHANG Z， et al. A hybrid trajectory planning strategy for intelligent vehicles in on-road dynamic scenarios［J］. IEEE Transactions on Vehicular Technology， 2022， 72（3）： 2832-2847.
[15]	LI L， GAN J， JI X， et al. Dynamic driving risk potential field model under the connected and automated vehicles environment and its application in car-following modeling［J］. IEEE Transactions on Intelligent Transportation Systems， 2020， 23（1）： 122-141.
[16]	HAN J， ZHAO J， ZHU B， et al. Spatial-temporal risk field for intelligent connected vehicle in dynamic traffic and application in trajectory planning［J］. IEEE Transactions on Intelligent Transportation Systems， 2023， 24（3）： 2963-2975.
[17]	TAN H， LU G， LIU M. Risk field model of driving and its application in modeling car-following behavior［J］. IEEE Transactions on Intelligent Transportation Systems， 2021， 23（8）： 11605-11620.
[18]	BAO S， BOYLE L N. Age-related differences in visual scanning at median-divided highway intersections in rural areas［J］. Accident Analysis & Prevention， 2009， 41（1）： 146-152.
[19]	AKAGI Y， RAKSINCHAROENSAK P. Stochastic driver speed control behavior modeling in urban intersections using risk potential-based motion planning framework［C］.2015 IEEE Intelligent Vehicles Symposium （IV）. IEEE， 2015： 368-373.
[20]	HAN Y， LI Q， QIAN Y， et al. Comparison of the landing kinematics of pedestrians and cyclists during ground impact determined from vehicle collision video records［J］. International Journal of Vehicle Safety， 2018， 10（3-4）： 212-234.
[21]	公安部. 道路交通信号控制系统术语： GB/T 31418—2015［S］. 2015.
	Ministry of Public Security. Road traffic signal control system terminology： GB/T 31418—2015 ［S］. 2015.
[22]	吴贺，韩勇，石亮亮，等. 基于视频信息的高精度事故重建方法研究［J］. 汽车工程， 2020， 42（6）： 778-783，792.
	WU H， HAN Y， SHI L， et al. Research on high-precision accident reconstruction method based on video information［J］. Automotive Engineering， 2020， 42（6）： 778-783，792.
[23]	金钱钱，韩勇，黄红武，等. 基于视频信息的电动两轮车在不同碰撞场景下TTC研究［C］.第22届中国汽车安全技术学术会议论文集. 保定：中国汽车工程学会安全技术分会， 2019.
	JIN Q， HAN Y， HUANG H， et al. TTC study of electric two-wheeled vehicles in different collision scenarios based on video information［C］. Proceedings of the 22nd China Automotive Safety Technology Conference. Baoding： Safety Technology Branch of China Society of Automotive Engineers， 2019.
[24]	NGUYEN H H. Privacy-preserving mechanisms for k-modes clustering［J］. Computers & Security， 2018， 78： 60-75.
[25]	SAMMOUDA R， EL-ZAART A. An optimized approach for prostate image segmentation using K‐means clustering algorithm with elbow method［J］. Computational Intelligence and Neuroscience， 2021， 2021（1）： 4553832.
[26]	UHM T， YI S. A comparison of normality testing methods by empirical power and distribution of P-values［J］. Communications in Statistics-Simulation and Computation， 2023， 52（9）： 4445-4458.
[27]	HIGGINS J J. An introduction to modern nonparametric statistics［M］. Pacific Grove， CA： Brooks/Cole， 2004.
[28]	BERGER P D， MAURER R E， CELLI G B. Experimental design：with applications in management， engineering， and the sciences［M］. 2nd ed. Cham： Springer， 2018.
[29]	CONOVER W J. Practical nonparametric statistics［M］. John Wiley & Sons， 1999.
[30]	LEHMANN E L. Elements of large-sample theory［M］. New York， NY： Springer New York， 1999.
[31]	LI Y， LI K， ZHENG Y， et al. Threat assessment techniques in intelligent vehicles： a comparative survey［J］. IEEE Intelligent Transportation Systems Magazine， 2020， 13（4）： 71-91.
[32]	MULLAKKAL-BABU F A， WANG M， HE X， et al. Probabilistic field approach for motorway driving risk assessment［J］. Transportation Research Part C： Emerging Technologies， 2020， 118： 102716.
[33]	American Psychological Association. Publication manual of the American psychological association （2020）［J］. American Psychological Association， 2019， 428.
[34]	ANDERSON D R， WILLIAMS T A， COCHRAN J J. Statistics for business & economics［M］. Cengage Learning， 2020.

特征名称		特征名称缩写	特征类型
车辆与障碍物运动学特征	车辆速度	VSpd	连续变量
	车辆加速度	VAcc	连续变量
	障碍物速度	OSpd	连续变量
	障碍物加速度	OAcc	连续变量
	碰撞时间	TTC	连续变量
道路环境特征	信号灯状态	SigS	有序分类变量
	有无分隔栏	DivAvl	二分类变量
	有无人行横道	CWAvl	二分类变量
	车道数	OWLNum	有序分类变量

信号灯状态	状态代码	状态详情
无信号灯	0	十字路口无交通信号灯
通行信号状态	1	十字路口存在交通信号灯，车辆正常通行，VRU横穿路口处于危险状态
警示信号状态	2	路口交通信号灯处于信号变换阶段，车辆与VRUs发生冲突

道路环境特征	Cluster 0 vs 1	Cluster 0 vs 2	Cluster 1 vs 2
道路环境特征	P*	P	P
SigS	<0.001	<0.001	<0.001
DivAvl	<0.001	<0.001	<0.001
CWAvl	<0.001	0.558	<0.001
OWLNum	<0.001	0.001	<0.001

运动学参数	场景1	场景2	场景3
运动学参数	P*	P	P
VSpd	<0.001	0.023	0.148
VAcc	0.022	<0.001	<0.001
OSpd	<0.001	<0.001	0.055
OAcc	<0.001	<0.001	0.190
TTC	<0.001	<0.001	0.025

运动学参数	CWAvl	DivAvl	SigS	OWLNum
运动学参数	P*	P	P	P
VSpd	0.025	<0.905	0.870	<0.001
VAcc	0.129	<0.921	<0.001	<0.001
OSpd	<0.001	<0.001	0.007	<0.001
OAcc	0.063	0.028	0.672	0.018
TTC	0.558	0.096	0.594	0.202