Complexity evaluation of vehicle-vehicle accident scenarios for autonomous driving simulation tests

doi:10.3969/j.issn.1674-8484.2022.04.011

Abstract

Abstract:

The complexity of vehicle-vehicle accident scenarios was evaluated to solve the problem of lack of basis for selecting autonomous driving simulation test scenarios. 670 vehicle-vehicle accidents were selected from the National Automobile Accident In-depth Investigation System (NAIS) database of China. 13 variables were extracted according to different dimensions of the scenario. The complexity evaluation model of vehicle-vehicle accident scenarios was established based on information entropy theory. The logical regression was used to obtain the variable level odds ratio to calculate the level complexity, and the dimensions and variables weights of the scenario were obtained by Back Propagation (BP) neural network algorithms. The model was applied to calculate the complexity of the scenario for each accident case. The scenario complexity was divided into four levels by K-means clustering. The dominant characteristics and deaths of various scenarios were obtained. The results show that level 4 complexity scenarios account for 1.6 %, but with a mortality rate of 90.9 %, such scenarios deserve priority attention

Key words: autonomous driving, vehicle-vehicle accidents, accident scenarios, complexity, information entropy

CLC Number:

U467

LI Pingfei, JIN Siyu, HU Wenhao, GAO Li, CHE Yaoyu, TAN Zhengping, DONG Xiaofei. Complexity evaluation of vehicle-vehicle accident scenarios for autonomous driving simulation tests[J]. Journal of Automotive Safety and Energy, 2022, 13(4): 697-704.

Figures/Tables 11

References 23

[1]	Menzel T, Bagschik G, Maurer M. Scenarios for development, test and validation of automated vehicles[C]// 2018 IEEE Intell Vehi Symp (IV). Changshu, China: IEEE, 2018: 1821-1827.
[2]	Koopman P. Challenges in autonomous vehicle validation: Keynote presentation abstract[C/OL]// Proc 1st Int’l Workshop Safe Control Connected Autonomous Vehi, New York: ACM, 2017. (2017-04-18). http://dl.acm.org/doi/10.1145/3055378.3055379.
[3]	谭正平, 何霞, 张道文, 等. 基于多信息融合的多车事故碰撞过程分析方法[J]. 中国司法鉴定, 2021(1): 98-104.
	TAN Zhengping, HE Xia, ZHANG Daowen, et al. Multi-vehicle accident collision process analysis method based on multi-information fusion[J]. *Chin J Forensic Sci*, 2021(1): 98-104. (in Chinese)
[4]	王清平, 钱宇彬, 冯浩, 等. 轿车与行人事故重建的参数敏感性研究[J]. 中国安全科学学报, 2019, 29(6): 70-75. doi: 10.16265/j.cnki.issn1003-3033.2019.06.012
	WANG Qingping, QIAN Yubin, FENG Hao, et al. Research on parameter sensitivity of car and pedestrian accident reconstruction[J]. *Chin Safe Sci J*, 2019, 29(6): 70-75. (in Chinese)
[5]	LIU Yu, WAN Xinming, XU Wei, et al. An intelligent method for accident reconstruction involving car and e-bike coupling automatic simulation and multi-objective optimizations[J]. *Accid Anal Prev, 2022, 164*: Paper No. 106476.
[6]	SUI Bo, Lubbe N, Bärgman J. A clustering approach to developing car-to-two-wheeler test scenarios for the assessment of Automated Emergency Braking in China using in-depth Chinese crash data[J]. *Accid Anal Prev, 2019, 132*: Paper No. 105242.
[7]	HU Wenhao, XU Xiangyang, ZHOU Zhaohui, et al. Mining and comparative analysis of typical pre-crash scenarios from IGLAD[J]. *Accid Anal Prev, 2020, 145*: Paper No. 105699.
[8]	TAN Zhengping, CHE Yaoyue, XIAO Lingyun, et al. Research of fatal car-to-pedestrian precrash scenarios for the testing of the active safety system in China[J]. *Accid Anal Prev, 2021, 150*: Paper No. 105857.
[9]	陈吉清, 舒孝雄, 兰凤崇, 等. 典型危险事故特征的自动驾驶测试场景构建[J]. 华南理工大学学报(自然科学版), 2021, 49(5): 1-8.
	CHEN Jiqing, SHU Xiaoxiong, LAN Fengchong, et al. Construction of automatic driving test scenarios with typical characteristics of dangerous accidents[J]. J South Chin Univ Tech (Sci and Tech), 2021, 49(5): 1-8. (in Chinese)
[10]	张海潮. 基于引力模型的道路交通环境复杂度研究[D]. 北京: 北京理工大学, 2016.
	ZHANG Haichao. Research on complexity of road traffic environment based on gravity model[D]. Bejing: Bejing Univ Tech, 2016. (in Chinese)
[11]	毕蕊. 基于脑电数据的交通环境复杂度辨识及风险预警[D]. 北京: 北方工业大学, 2020.
	BI Rui. The traffic environment complexity identification and risk warning based on EEG data[D]. Bejing: Beifang Univ Tech, 2020. (in Chinese)
[12]	张朋. 道路动态交通环境复杂度量化研究[D]. 长春: 吉林大学, 2020.
	ZHANG Peng. Quantitative study on complexity of road dynamic traffic environment[D]. Changchun: Jilin Univ, 2020. (in Chinese)
[13]	王宇雷, 李凯, 胡云峰, 等. 一种面向复杂超车场景的行驶任务复杂度量化评估方法与流程: 中国, CN110182217A[P]. 2019.
	WANG Yulei, LI Kai, HU Yunfeng, et al. A quantitative evaluation method and process for driving task complexity in complex overtaking scenarios: China, CN110182217A[P]. 2019. (in Chinese)
[14]	董汉, 舒伟, 陈超, 等. 危险驾驶工况场景的复杂度评估方法研究[J]. 汽车工程, 2020, 42(6): 808-814.
	DONG Han, SHU Wei, CHEN Chao, et al. Dangerous driving conditions scenario complexity evaluation method research[J]. *Auto Engi*, 2020, 42(6): 808-814. (in Chinese)
[15]	李江坤, 邓伟文, 任秉韬, 等. 一种智能汽车测试场景复杂度的评估方法[C]// 2020中国汽车工程学会年会论文集 (1), 2020: 110-117.
	LI Jiangkun, DENG Weiwen, REN Bingtao, et al. An evaluation method for the complexity of intelligent vehicle test scenarios[C]// 2020 Chin Ass Auto Engi Annu Conf Papers (1), 2020: 110-117. (in Chinese)
[16]	王荣, 孙亚夫, 宋娟. 自动驾驶车辆道路测试场景评价方法与试验验证[J]. 汽车工程, 2021, 43(4): 620-628.
	WANG Rong, SUN Yafu, SONG Juan. Evaluation method and test verification of road test scenarios for autonomous vehicles[J]. *Auto Engi*, 2021, 43(4): 620-628. (in Chinese)
[17]	曹毅, 周华, 肖凌云, 等. 基于NAIS数据库中视频信息的人—车碰撞事故特征分析[J]. 汽车安全与节能学报, 2020, 11(1): 44-52.
	CAO Yi, ZHOU Hua, XIAO Linyun, et al. Analysis of pedestrian-vehicle collision accident characteristics based on the video information from NAIS database[J]. *J Auto Safe Energy*, 2020, 11(1): 44-52. (in Chinese)
[18]	高岩, 苏虎, 于洋, 等. 智能车辆仿真场景建模方法[J]. 交通信息与安全, 2020, 38(1): 100-106.
	GAO Yan, SU Hu, YU Yang, et al. A modeling method of intelligent vehicle simulation scenario[J]. *J Trans Info Safe*, 2020, 38(1): 100-106. (in Chinese)
[19]	朱波, 胡旭东, 谈东奎, 等. 基于多通道态势图的自动驾驶场景表征方法[J]. 中国公路学报, 2020, 33(8): 204-214. doi: 10.19721/j.cnki.1001-7372.2020.08.020
	ZHU Bo, HU Xudong, TAN Dongkui, et al. Automatic driving scenario representation method based on multi-channel situation map[J]. *Chin J H-way Trans*, 2020, 33(8): 204-214. (in Chinese)
[20]	冯屹, 王兆. 自动驾驶测试场景技术发展与应用[M]. 北京: 机械工业出版社, 2020: 125-126.
	FENG Yi, WANG Zhao. Development and Application of Autonomous Driving Test Scenarios[M]. Bejing: China Machine Press, 2020: 125-126. (in Chinese)
[21]	王磊, 吕璞, 林永杰. 高速公路交通事故影响因素分析及伤害估计[J]. 中国安全科学学报, 2016, 26(3): 86-90.
	WANG Lei, LÜ Pu, LIN Yongjie. Traffic accidents on freeways: Influencing factors analysis and injury severity evaluation[J]. *Chin Safe Sci J*, 2016, 26(3): 86-90. (in Chinese)
[22]	孟瑾. 因子分析和神经网络的信息系统风险评估模型[J]. 现代电子技术, 2020, 43(23): 62-66.
	MENG Jin. Information system risk assessment model based on factor analysis and neural network[J]. *Modern Electr Tech*, 2020, 43(23): 62-66. (in Chinese)
[23]	王林泽, 高德东, 白辉全, 等. 基于力视感知的斜尖柔性针轨迹预测[J]. 机械工程学报, 2021, 57(11): 52-60. doi: 10.3901/JME.2021.11.052
	WANG Linze, GAO Dedong, BAI Huiquan, et al. The trajectory prediction of oblique-tip flexible needle based on force visual perception[J]. *J Mech Engi*, 2021, 57(11): 52-60. (in Chinese)

采集指标	指标水平	占比/%
采集指标	指标水平	NAIS数据库	统计年报	差值
天气	晴或多云	70.52	74.75	4.23
	阴	15.04	14.47	0.57
	雨天等恶劣天气	14.10	10.78	3.32
时段	日间	45.93	58.12	12.19
	晨昏	9.59	4.57	5.02
	夜晚	44.49	37.30	7.19
道路类型	普通路段	49.96	73.30	23.34
	路口	36.64	20.12	16.52
	其他	13.40	6.58	6.82
事故类型	机动车与机动车	40.17	69.23	9.96
	机动车与两轮车	19.10	69.23	9.96
	机动车碰撞行人	15.76	22.94	7.18
	机动车本车事故	24.97	7.83	17.14

采集指标	指标水平	占比/%
采集指标	指标水平	NAIS数据库	统计年报	差值
天气	晴或多云	70.52	74.75	4.23
	阴	15.04	14.47	0.57
	雨天等恶劣天气	14.10	10.78	3.32
时段	日间	45.93	58.12	12.19
	晨昏	9.59	4.57	5.02
	夜晚	44.49	37.30	7.19
道路类型	普通路段	49.96	73.30	23.34
	路口	36.64	20.12	16.52
	其他	13.40	6.58	6.82
事故类型	机动车与机动车	40.17	69.23	9.96
	机动车与两轮车	19.10	69.23	9.96
	机动车碰撞行人	15.76	22.94	7.18
	机动车本车事故	24.97	7.83	17.14

变量	V	L = 1	L = 2	L = 3	L = 4	L = 5	L = 6	L = 7
主车碰撞速度/ (km·h^-1)	1	0~20	20~40	40~60	60~80	80~100	>100
主车运动状态	2	直行	左转	右转	向左变道	向右变道	左转弯掉头	停车
视野遮挡	3	车辆造成	其他物体造成	无视野遮挡
目标车类型	4	乘用车	微型客和货车	轻型客和货车	中型客和货车	重型客和货车
目标车碰撞速度/ (km·h^-1)	5	0~20	20~40	40~60	60~80	80~100	>100
目标车运动状态	6	直行	左转	右转	向左变道	向右变道	左转弯掉头	停车
相对位置	7	左方来车	右方来车	对向	同向
道路类型	8	普通路段	路口	路段进出口	隧道	左弯道	右弯道	匝道
道路行政等级	9	一级公路	二级公路	三级公路	乡村道路	城市道路	厂矿道路	高速公路
信号灯	10	直行+转向	仅直行	无信号灯
天气	11	晴	阴	多云	雨	雪	雾
时段	12	日间	夜间	晨昏
路灯	13	无	有且点亮	有但关闭

变量	V	L = 1	L = 2	L = 3	L = 4	L = 5	L = 6	L = 7
主车碰撞速度/ (km·h^-1)	1	0~20	20~40	40~60	60~80	80~100	>100
主车运动状态	2	直行	左转	右转	向左变道	向右变道	左转弯掉头	停车
视野遮挡	3	车辆造成	其他物体造成	无视野遮挡
目标车类型	4	乘用车	微型客和货车	轻型客和货车	中型客和货车	重型客和货车
目标车碰撞速度/ (km·h^-1)	5	0~20	20~40	40~60	60~80	80~100	>100
目标车运动状态	6	直行	左转	右转	向左变道	向右变道	左转弯掉头	停车
相对位置	7	左方来车	右方来车	对向	同向
道路类型	8	普通路段	路口	路段进出口	隧道	左弯道	右弯道	匝道
道路行政等级	9	一级公路	二级公路	三级公路	乡村道路	城市道路	厂矿道路	高速公路
信号灯	10	直行+转向	仅直行	无信号灯
天气	11	晴	阴	多云	雨	雪	雾
时段	12	日间	夜间	晨昏
路灯	13	无	有且点亮	有但关闭

检验		卡方	自由度	显著性
Omnibus	步骤	338.818	54	0.000
	块	338.818	54	0.000
	模型	338.818	54	0.000
Hosmer Lemeshow	步骤	7.788	8	0.454