基于IAHP-EWM-LDM的L3级自动驾驶接管过程安全性评价方法

李振龙; 潘梦妞; 曲彦菘; 赵晓华; 巩建国; 王秋鸿

doi:10.3963/j.jssn.1674-4861.2023.04.002

基于IAHP-EWM-LDM的L3级自动驾驶接管过程安全性评价方法

doi: 10.3963/j.jssn.1674-4861.2023.04.002

1.
北京工业大学城市交通学院北京 100124
2.
公安部道路交通安全研究中心北京 100062

基金项目:

公安理论及软科学研究计划项目 2021LL62

详细信息

通讯作者:
李振龙（1976—），博士，教授.研究方向: 交通控制、驾驶行为等. E-mail: lzl@bjut.edu.cn

中图分类号: U491.2+5
计量
- 文章访问数: 551
- HTML全文浏览量: 255
- PDF下载量: 41
- 被引次数: 0
出版历程
- 收稿日期: 2022-07-21
- 网络出版日期: 2023-11-23

A Method for Evaluating the Safety over the Takeover Process of the Level 3 Automated Vehicles Based on IAHP-EWM-LDM

1.
College of Metropolitan Transportation, Beijing University of Technology, Beijing 100124, China
2.
Research Institute for Road Safety of Ministry of Public Security, Beijing 100062, China

摘要

摘要: 在L3级自动驾驶阶段，驾驶人需要在系统发出接管请求时，及时响应并接管车辆。因此为了准确评估L3级自动驾驶车辆接管过程的安全性，构建了自动驾驶接管过程的安全性评价指标体系。本文采用4×2×2的接管场景因子设计了驾驶模拟试验，利用驾驶模拟器采集各类驾驶数据；基于变异系数法和Spearman相关性判别法从风险感知、避险操纵和接管绩效3个方面分析得到了13个安全性评价指标；使用能够表征专家经验的改进层次分析法求取指标的主观权重，使用能够反映数据特征的熵权法求取指标的客观权重；为综合2种方法的优点，利用级差最大化法获得了融合主、客观权重的综合权重，并通过计算得出风险感知、避险操纵、接管绩效的综合权重分别为0.259、0.475、0.271，以此结果来构建接管过程的安全性评价指标体系。本文运用该体系对驾驶模拟试验所获得的655个接管过程进行了综合评价，根据评价结果将其划分为A、B、C这3类接管过程。对比3类接管过程在风险感知、避险操纵、接管绩效3个方面的得分发现，A类接管过程在3个方面均表现较好，C类接管过程在避险操纵和接管绩效2个方面表现较差，B类指标的表现介于A类和C类之间，不同类别的接管过程在各个指标上均具有较好的区分度。本文构建的评价指标体系有效结合了专家经验和指标特征，能够为更加全面、合理和科学地评价自动驾驶接管过程中的安全性提供理论支持。
- 交通安全 /
- 接管过程 /
- 评价指标体系 /
- 改进层次分析法 /
- 熵权法 /
- 级差最大化法
Abstract: In the Level 3 autonomous driving stage, the driver needs to respond and take over the vehicle when the system sends a takeover request. Therefore, to accurately assess the safety of the takeover process of Level 3 autonomous vehicles, the safety evaluation index system of the takeover process of autonomous driving is constructed. In this paper, a 4×2×2 takeover scenario factor is used to design a driving simulation test, and a driving simulator is used to collect various types of driving data. Based on the coefficient of variation method and Spearman correlation discriminant method, 13 security evaluation indicators are obtained from the analysis of 3 aspects, such as risk perception, risk avoidance manipulation and takeover performance. The subjective weights of the indicators are obtained using an improved hierarchical analysis that characterizes the experience of the experts, and the subjective weights of the indicators are obtained using entropy weights that reflect the characteristics of the data. To combine the advantages of the two methods, a composite weight incorporating both subjective and objective weights is obtained using the grade maximization method. The combined weights of risk perception, risk avoidance manipulation, and takeover performance are calculated to be 0.259, 0.475, and 0.271, which are used to construct the security evaluation index system of the takeover process. In this paper, the system is applied to comprehensively evaluate 655 takeover processes obtained from driving simulation tests, and they are classified into 3 categories of A, B and C takeover processes according to the evaluation results. Comparing the scores of the 3 types of takeover processes in 3 aspects: risk perception, risk avoidance manipulation and takeover performance, it is found that the A-type takeover process performs well in three aspects, the C-type takeover process performs poorly in risk avoidance manipulation and takeover performance, and the B-type takeover process performs intermediary between the A-type and C-type. Different types of takeover process have a better degree of differentiation in each indicator. The indicator system is constructed that effectively combines expert experience and indicator characteristics. The evaluation index system constructed in this paper effectively combines expert experience and index characteristics. It can provide theoretical support for a more comprehensive, reasonable and scientific evaluation of the safety in the process of automatic driving takeover.
- traffic safety /
- takeover process /
- evaluation indicators system /
- improved analytic hierarchy process /
- entropy weight method /
- level difference maximization

HTML全文

图 1 试验设备

Figure 1. The laboratory equipment

下载: 全尺寸图片幻灯片

图 2 接管情境示意图

Figure 2. Schematic diagram of takeover scenario

下载: 全尺寸图片幻灯片

图 3 接管场景布设

Figure 3. Location of takeover scenarios

下载: 全尺寸图片幻灯片

图 4 接管过程安全性评价指标体系

Figure 4. The takeover process safety evaluation indicators system

下载: 全尺寸图片幻灯片

图 5 最优和最差接管过程的归一化指标

Figure 5. Best and worst takover process normalized indicators

下载: 全尺寸图片幻灯片

表 1 驾驶人信息

Table 1. Driver information

	驾驶人属性	数量/人	平均值	标准差
性别	男	32
性别	女	10
年龄/岁	青年（≥18~35）	15	23.2	2.0
	中年（> 35~60）	14	46.5	6.6
	老年（> 60）	13	63.7	2.9
驾龄/年	低驾龄（≥1~12）	18	4.1	2.7
	中驾龄（> 12~26）	14	20.2	4.7
	高驾龄（> 26）	10	32.0	4.4

下载: 导出CSV

表 2 接管情景组合及编号

Table 2. Takeover scenario combinations and numbers

接管情境	驾驶次任务	接管请求时间/s	接管场景
主线情境	看视频答题	5	S1
	看视频答题	10	S2
	微信发语音	5	S3
	微信发语音	10	S4
团雾情境	看视频答题	5	S5
	看视频答题	10	S6
	微信发语音	5	S7
	微信发语音	10	S8
事故情境	看视频答题	5	S9
	看视频答题	10	S10
	微信发语音	5	S11
	微信发语音	10	S12
匝道情境	看视频答题	5	S13
	看视频答题	10	S14
	微信发语音	5	S15
	微信发语音	10	S16

下载: 导出CSV

表 3 HMI界面

Table 3. HMI interface

自动驾驶状态	描述	系统状态显示
不可用-H1	周围环境不满足自动驾驶要求，自动驾驶系统不可用
可用-H2	自动驾驶系统尚未激活，可按下“启动”按钮激活
已激活-H3	自动驾驶系统已激活，自动驾驶系统可用
接管请求-H4	系统发出请求，驾驶人按下“接管”按钮，控制权切换为驾驶人

下载: 导出CSV

表 4 初选参数的变异系数

Table 4. The coefficient variation of primary parameters

序号	参数	均值	标准差	变异系数/%	KS检验
1	Fix_time	0.250	0.131	52.400	0.05
2	Sac_time	0.111	0.153	137.838	0.05
3	Rate_pupil	0.029	0.020	68.966	0.05
4	Diff_pupil	22.239	12.989	58.406	0.05
5	Mani_time	6.515	7.454	114.413	0.05
6	Mani_TTC	5.151	4.418	85.77	0.05
7	Std_brake	16.612	24.314	146.364	0.05
8	Std_wheel	0.110	0.162	147.273	0.05
9	Std_gas	0.085	0.058	68.235	0.05
10	Std_a_x	0.352	0.285	80.966	0.05
11	Std_lp	0.612	0.264	43.137	0.05
12	Std_a_y	0.831	0.689	82.912	0.05
13	Std_speed	11.213	6.836	60.965	0.05

下载: 导出CSV

表 5 初选参数相关系数（部分）

Table 5. The correlation coefficient of primary parameters (part)

相关系数	Sac_time	Diff_pupil	Mani_time	Std_gas
Fix_time	0.053	0.152	0.063	0.070
Rate_pupil	0.370	0.146	-0.015	-0.010
Mani_TTC	-0.003	0.038	-0.348	0.087
Std_wheel	-0.072	-0.216	0.016	0.265
Std_a_x	-0.129	-0.297	-0.035	0.137

下载: 导出CSV

表 6 指标权重及差异情况

Table 6. The index weight and differences

指标	均值	标准差	变异系数/%
Fix_time	0.090	0.012	13.333
Sac_time	0.090	0.013	14.444
Rate_pupil	0.085	0.010	11.765
Diff_pupil	0.082	0.008	9.756
Mani_time	0.087	0.011	12.644
Mani_TTC	0.086	0.015	17.442
Std_brake	0.058	0.010	17.241
Std_wheel	0.060	0.007	11.667
Std_gas	0.052	0.011	21.154
Std_a_x	0.078	0.012	15.385
Std_lp	0.080	0.014	17.500
Std_a_y	0.076	0.011	14.474
Std_speed	0.076	0.015	19.737

下载: 导出CSV

表 7 评价指标权重

Table 7. The evaluation index weight

一级指标	二级指标	三级指标	权重
一级指标	二级指标	三级指标	IAHP	EWM	LDM
风险感知	视觉感知	Fix_time	0.090	0.063	0.063
	视觉感知	Sac_time	0.090	0.041	0.041
	视觉负荷	Rate_pupil	0.085	0.075	0.075
	视觉负荷	Diff_pupil	0.082	0.050	0.080
避险操纵	接管反应	Mani_time	0.087	0.029	0.079
	接管反应	Mani_TTC	0.086	0.124	0.116
	执行阶段	Std_brake	0.058	0.168	0.108
		Std_wheel	0.060	0.129	0.120
		Std_gas	0.052	0.071	0.052
接管绩效	横向表现	Std_a_x	0.078	0.097	0.078
	横向表现	Std_lp	0.080	0.032	0.072
	纵向表现	Std_a_y	0.076	0.067	0.067
	纵向表现	Std_speed	0.076	0.054	0.054

下载: 导出CSV

表 8 指标权重偏差

Table 8. Index weight deviation

指标	平均值	IAHP/%	EWM/%	LDM/%
Fix_time	0.072	26	13	13
Sac_time	0.057	57	29	29
Rate_pupil	0.078	9	4	4
Diff_pupil	0.071	16	29	13
Mani_time	0.065	33	55	22
Mani_TTC	0.109	21	14	7
Std_brake	0.111	48	51	3
Std_wheel	0.103	42	25	16
Std_gas	0.058	11	22	11
Std_a_x	0.084	8	15	8
Std_lp	0.061	31	48	17
Std_a_y	0.070	9	4	4
Std_speed	0.062	23	12	12
累计偏差		332	321	159

下载: 导出CSV

表 9 3类接管过程评价指标平均值

Table 9. Average of three types of takeover process evaluation indicators

一级指标	指标	指标序号	A	B	C
风险感知	Fix_time	1	0.24	0.25	0.27
	Sac_time	2	0.11	0.11	0.11
	Rate_pupil	3	0.02	0.03	0.03
	Diff_pupil	4	17.66	22.25	23.46
避险操纵	Mani_time	5	5.77	6.91	6.45
	Mani_TTC	6	8.52	4.25	3.64
	Std_brake	7	5.57	11.29	38.29
	Std_wheel	8	0.05	0.07	0.25
	Std_gas	9	0.06	0.08	0.13
接管绩效	Std_a_x	10	0.27	0.34	0.46
	Std_lp	11	0.53	0.59	0.74
	Std_a_y	12	0.43	0.65	1.60
	Std_speed	13	7.10	10.19	17.44

下载: 导出CSV

表 10 接管过程评价指标对比

Table 10. Comparison of evaluation indicators of the takeover process

一级指标	指标	R_AB /%	R_BC /%
风险感知	Fix_time	95.12	61.89
	Sac_time	10.00	13.90
	Rate_pupil	87.20	1.83
	Diff_pupil	18.90	4.57
避险操纵	Mani_time	28.66	15.24
	Mani_TTC	63.41	10.98
	Std_brake	74.39	86.89
	Std_wheel	87.20	29.88
	Std_gas	70.73	57.93
接管绩效	Std_a_x	75.61	82.62
	Std_lp	76.22	94.82
	Std_a_y	66.46	58.84
	Std_speed	73.17	35.37

下载: 导出CSV

表 11 G₁₅和 G₂₂₉的评价指标

Table 11. The evaluation indicators of G₁₅ and G₂₂₉

一级指标	指标	指标序号	G₂₉₉	G₁₅
风险感知	Fix_time	1	0.16	0.22
	Sac_time	2	0.05	0.33
	Rate_pupil	3	0.02	0.04
	Diff_pupil	4	4.91	13.10
避险操纵	Mani_time	5	5.50	5.92
	Mani_TTC	6	16.14	2.50
	Std_brake	7	0.00	63.04
	Std_wheel	8	0.12	0.32
	Std_gas	9	0.01	0.27
接管绩效	Std_a_x	10	0.45	1.19
	Std_lp	11	0.64	0.83
	Std_a_y	12	0.10	3.62
	Std_speed	13	2.98	22.82

下载: 导出CSV

参考文献(26)

[1]	SAE International. Taxonomy and definitions for terms related to on-road motor vehicle automated driving systems: SAE J3016-2021[S]. Warrendale: SAE International, 2021.
[2]	中华人民共和国工业和信息化部. 汽车驾驶自动化分级: GB/T 40429—2021[S]. 北京: 中国标准出版社, 2021. Ministry of Industry and Information Technology of the People's Republic of China. Taxonomy of drivingautomation for Vehicles: GB/T 40429—2021[S]. Beijing: China Standard Press, 2021. (in Chinese)
[3]	GOOGLE. Google self-driving car testing report on disengagements of autonomous mode[R]. Google Auto LLC, 2015, 92(12): 249-255.
[4]	鲁光泉, 陈发城, 李鹏辉, 等. 驾驶人跟车风险接受水平对其接管绩效的影响[J]. 汽车工程, 2021, 43(6): 808-814. doi: 10.19562/j.chinasae.qcgc.2021.06.003 LU G Q, CHEN F C, LI P H, et al. Effect of drivers' acceptance level of car-following risk on the takeover performance[J]. Automotive Engineering, 2021, 43(6): 808-814. (in Chinese) doi: 10.19562/j.chinasae.qcgc.2021.06.003
[5]	安玉, 焦朋朋, 白紫秀. 考虑多因素的驾驶行为安全评价与风险等级预测[J]. 系统仿真学报, 2021, 33(1): 118-126. https://www.cnki.com.cn/Article/CJFDTOTAL-XTFZ202101015.htm AN Y, JIAO P P, BAI Z X. Safety evaluation and risk level prediction of driving behavior considering multi-factors influence[J]. Journal of System Simulation, 2021, 33(1): 118-126. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-XTFZ202101015.htm
[6]	SCHÖNER H P, PRETTO P, SODNIK J, et al. A safety score for the assessment of driving style[J]. Traffic Injury Prevention, 2021, 22(5): 384-389. doi: 10.1080/15389588.2021.1904508
[7]	咸化彩, 金立生, 李科勇, 等. 基于网络分析法的次任务驾驶安全评价模型[J]. 长安大学学报(自然科学版), 2015, 35: 124-128. https://www.cnki.com.cn/Article/CJFDTOTAL-XAGL2015S1027.htm XIAN H C, JIN L S, LI K Y, et al. Driving safety evaluation model for secondary task driving by using the method of ANP[J]. Journal of Chang'an University(Nature Science Edition), 2015, 35: 124-128. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-XAGL2015S1027.htm
[8]	刘志强, 王涛. 基于改进TOPSIS的驾驶行为实时安全性评估方法[J]. 重庆理工大学学报(自然科学), 2021, 35(11): 58-66. https://www.cnki.com.cn/Article/CJFDTOTAL-CGGL202111008.htm LIU Z Q, WANG T. Real-time safety assessment method of driving behavior based on improved topsis[J]. Journal of Chongqing University of Technology(Natural Science), 2021, 35(11): 58-66. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-CGGL202111008.htm
[9]	LIN Q F, LI S Q, MA X W, et al. Understanding take-over performance of high crash risk drivers during conditionally automated driving[J]. Accident Analysis & Prevention, 2020, 143: 105543.
[10]	林庆峰, 王兆杰, 鲁光泉. L3级自动驾驶汽车的接管安全性评价模型[J]. 汽车工程, 2019, 41(11): 1258-1264. https://www.cnki.com.cn/Article/CJFDTOTAL-QCGC201911006.htm LIN Q F, WANG Z J, LU G Q. Takeover safety evaluation model for level 3 automated vehicles[J]. Automotive Engineering, 2019, 41(11): 1258-1264. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-QCGC201911006.htm
[11]	GOLD C, HAPPEE R, BENGLER K. Modeling take-over performance in level 3 conditionally automated vehicles[J]. Accident Analysis & Prevention, 2017, 116: 3-13.
[12]	钮建伟, 张雪梅, 孙一品, 等. 险情中驾驶人接管自动驾驶车辆的驾驶行为研究[J]. 中国公路学报, 2018, 31(6): 272-280. https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGL201806015.htm NIU J W, ZHANG X M, SUN Y P, et al. Analysis of the driving behavior during the takeover of automatic driving vehicles in dangerous traffic situations[J]. China Journal of Highway and Transport, 2018, 31(6): 272-280. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGL201806015.htm
[13]	马小翔, 陈丰, 张霖. 预期接管场景下接管绩效及接管风险研究[J]. 中国公路学报, 2022, 35(1): 159-168. https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGL202201014.htm MA X X, CHEN F, ZHANG L. Takeover-performance and takeover-risk evaluation under non-critical transition scenarios[J]. China Journal of Highway and Transport, 2022, 35 (1): 159-168. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGL202201014.htm
[14]	周荣贵, 钟连德. 公路通行能力手册[M]. 北京: 人民交通出版社, 2017. ZHOU R G, ZHONG L D. Highway capacity manual[M]. Beijing: China Communications Press, 2017. (in Chinese)
[15]	ZEEB K, BUCHNER A, SCHRAUF M. Is take-over time all that matters? The impact of visual-cognitive load on driver take-over quality after conditionally automated driving[J]. Accident Analysis & Prevention, 2016, 92: 230-239.
[16]	ITO T, TAKATA A, OOSAWA K. Time required for take-over from automated to manual driving[J]. SAE Technical Paper, 2016, 158(4): 1-6.
[17]	林庆峰, 王兆杰, 鲁光泉. 城市道路环境下自动驾驶车辆接管绩效分析[J]. 中国公路学报, 2019, 32(6): 240-247. https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGL201906025.htm LIN Q F, WANG Z J, LU G Q. Analysis of take-over performance for automated vehicles in urban road environments[J]. China Journal of Highway and Transport, 2019, 32(6): 240-247. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGL201906025.htm
[18]	冯忠祥, 李靖宇, 张卫华, 等. 面向人机共驾车辆的驾驶人风险感知研究综述[J]. 交通信息与安全, 2022, 40(2): 1-10. doi: 10.3963/j.jssn.1674-4861.2022.02.001 FENG Z X, LI J Y, ZHANG W H, et al. A reviewon driver's perception of risk associated with autonomous driving under human-computer shared control[J]. Journal of Transport Information and Safety, 2022, 40(2): 1-10. (in Chinese) doi: 10.3963/j.jssn.1674-4861.2022.02.001
[19]	郭应时. 交通环境及驾驶经验对驾驶员眼动和工作负荷影响的研究[D]. 西安: 长安大学, 2009. GUO Y S. Study on effects of traffic environment and driving experience on driver's eye movement and workload[D]. Xi'an: Chang'an University, 2009. (in Chinese)
[20]	郭子慧, 郭伟伟, 谭墍元. 驾驶员接管自动驾驶车辆的眼动特性和行为分析[J]. 中国安全科学学报, 2022, 32(1): 65-71. https://www.cnki.com.cn/Article/CJFDTOTAL-ZAQK202201009.htm GUO Z H, GUO W W, TAN X Y. Analysis on eye movement characteristics and behavior of drivers taking over automated vehicles[J]. China Safety Science Journal, 2022, 32 (1): 65-71. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZAQK202201009.htm
[21]	梅家林, 杜志刚, 郑号染, 等. 不同时段特长隧道入口区域视觉负荷研究[J]. 中国安全科学学报, 2021, 31(6): 176-181. https://www.cnki.com.cn/Article/CJFDTOTAL-ZAQK202106025.htm MEI J L, DU Z G, ZHENG H R, et al. Research on visual load at entrance area of extra-long tunnel in different periods[J]. China Safety Science Journal, 2021, 31(6): 176-181. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZAQK202106025.htm
[22]	鲁光泉, 赵鹏云, 王兆杰, 等. 自动驾驶中视觉次任务对年轻驾驶人接管时间的影响[J]. 中国公路学报, 2018, 31(4): 165-171. https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGL201804021.htm LU G Q, ZHAO P Y, WANG Z J, et al. Impact of visual secondary task on young drivers' take-over time in automated driving[J]. China Journal of Highway and Transport, 2018, 31(4): 165-171. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGL201804021.htm
[23]	李鹏辉, 廖呈玮, 郑志晓, 等. 认知分心对车辆跟驰过程操控安全性的影响[J]. 中国公路学报, 2018, 31(5): 167-173. https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGL201805021.htm LI P H, LIAO C W, ZHENG Z X, et al. Impact of cognitive distraction on vehicle control safety in car-following situation[J]. China Journal of Highway and Transport, 2018, 31 (5): 167-173. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGL201805021.htm
[24]	徐超. 基于虚拟驾驶的切换型人机共驾行为评价[D]. 长春: 吉林大学, 2018. XU C. Evaluation of cooperative driving behavior based on the virtual driving[D]. Changchun: Jilin University, 2018. (in Chinese)
[25]	赵微, 林健, 王树芳, 等. 变异系数法评价人类活动对地下水环境的影响[J]. 环境科学, 2013, 34(4): 1277-1283. https://www.cnki.com.cn/Article/CJFDTOTAL-HJKZ201304011.htm ZHAO W, LIN J, WANG S F, et al. Influence of human activities on groundwater environment based on coefficient variation method[J]. Environmental Science, 2013, 34(4): 1277-1283. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-HJKZ201304011.htm
[26]	HAO R, YANG H, ZHOU Z. Driving behavior evaluation model base on big data from internet of vehicles[J]. International Journal of Ambient Computing and Intelligence, 2019, 10(4): 78-95.