面向自动驾驶测试的危险变道场景泛化生成

赵祥模; 赵玉钰; 景首才; 惠飞; 刘建蓓

doi:10.16383/j.aas.c220772

面向自动驾驶测试的危险变道场景泛化生成

doi: 10.16383/j.aas.c220772

赵祥模^1,,
赵玉钰^1,,
景首才^{1, 2,},
惠飞^1,,
刘建蓓^{2, 3,}

1.
长安大学信息工程学院西安 710064
2.
中交第一公路勘察设计研究院有限公司西安 710075
3.
交通运输部交通安全与应急保障技术行业研发中心西安 710075

基金项目: 国家重点研发计划(2021YFB2501200)资助

详细信息

作者简介:
赵祥模：长安大学信息工程学院教授. 2006 年获得长安大学博士学位. 主要研究方向为交通信息技术与智慧交通, 智能网联汽车测试技术. E-mail: xmzhao@chd.edu.cn

赵玉钰：长安大学信息工程学院硕士研究生. 主要研究方向为智能网联汽车测试技术. E-mail: yuyuzhao@chd.edu.cn

景首才：长安大学信息工程学院讲师, 中交第一公路勘察设计研究院有限公司博士后. 分别于2014 年和2020 年获得长安大学自动化学士学位和交通信息工程与控制博士学位. 主要研究方向为智能网联车辆协同控制方法与测试技术. 本文通信作者. E-mail: scjing@chd.edu.cn

惠飞：长安大学信息工程学院教授. 2009 年获得西安微电子技术学院计算机系统架构系博士学位. 主要研究方向为网联车辆与图像处理. E-mail: feihui@chd.edu.cn

刘建蓓：中交第一公路勘察设计研究院有限公司教授级高级工程师. 主要研究方向为公路几何设计理论与方法, 交通安全评价与主动防控、保障, 智能交通控制与优化. E-mail: liujp09@gmail.com

计量
- 文章访问数: 622
- HTML全文浏览量: 343
- PDF下载量: 212
- 被引次数: 0
出版历程
- 收稿日期: 2022-09-27
- 录用日期: 2023-01-16
- 网络出版日期: 2023-09-21
- 刊出日期: 2023-10-24

Generalization Generation of Hazardous Lane-changing Scenarios for Automated Vehicle Testing

ZHAO Xiang-Mo^1
,,
ZHAO Yu-Yu^1
,,
JING Shou-Cai^{1, 2
,},
HUI Fei^1
,,
LIU Jian-Bei^{2, 3
,}

1.
School of Information Engineering, Chang＇an University, Xi＇an 710064
2.
China Communications Construction Company First Highway Consultants Limited Company, Xi＇an 710075
3.
Research and Development Center of Traffic Safety and Emergency Security Technology Industry of the Ministry of Transport, Xi＇an 710075

Funds: Supported by National Key Research and Development Program of China (2021YFB2501200)

More Information

Author Bio:
ZHAO Xiang-Mo　Professor at the School of Information Engineering, Chang＇an University. He received his Ph.D. degree from Chang＇an University in 2006. His research interest covers transportation information technology and intelligent transportation, intelligent and connected vehicle test technology

ZHAO Yu-Yu　Master student at the School of Information Engineering, Chang＇an University. Her main research interest is intelligent and connected vehicle test technology

JING Shou-Cai　Lecturer at the School of Information Engineering, Chang＇an University. Postdoctor in China Communications Construction Company First Highway Consultants Limited Company. He received his bachelor degree in automation and Ph.D. degree in traffic information engineering and control from Chang＇an University in 2014 and 2020, respectively. His main research interest is intelligent and connected vehicle cooperative control and test technology. Corresponding author of this paper

HUI Fei　Professor at the School of Information Engineering, Chang＇an University. He received his Ph.D. degree in Department of Computer System Architecture from Xi＇an Institute of Microelectronics Technology in 2009. His research interest covers connected vehicles and image processing

LIU Jian-Bei　Professor-level senior engineer of China Communications Construction Company First Highway Consultants Limited Company. Her research interest covers highway geometric design theory and method, traffic safety evaluation and active prevention and control, intelligent traffic control and optimization

摘要

摘要: 针对自动驾驶虚拟测试中危险变道场景构建问题, 提出一种数据−模型驱动的自动驾驶测试危险变道场景泛化生成方法. 基于 NGSIM US101 数据集中的紧急变道数据, 提出一种紧急变道轨迹对抗生成方法(BN-AM-SeqGAN), 构建基于安全距离的两车变道状态约束模型, 设计危险变道测试场景泛化生成方法, 生成危险变道测试场景库. 实验结果显示: 生成的5万条紧急变道轨迹变道完成时间分布的均方根误差为 0.63, 生成的 5 万个危险变道场景中, 99.54% 的场景被测自动驾驶车辆与变道背景车辆的碰撞时间小于 1 s, 表明该方法能够有效生成自动驾驶测试危险变道场景.
- 智能车辆 /
- 自动驾驶测试 /
- 变道场景 /
- 危险场景生成 /
- 序列生成对抗网络
Abstract: To address the issue of hazardous lane-changing scenario construction in automated vehicle virtual testing, proposed a data-model-driven method for generally producing hazardous lane-changing scenarios. Based on emergency lane-changing data in NGSIM US101 Dataset, an emergency lane-changing trajectories producing method called batch normalization-attention mechanism-sequence generative adversarial nets (BN-AM-SeqGAN) with policy gradient is proposed based on sequence generative adversarial network. The safety distance based constraint model for two vehicle lane-changing statesis built, and the general approach of producing hazardous lane-changing test scenarios is designed. The library of hazardous lane-changing test scenarios is finally achieved. According to the experimental findings, the root mean square error of the lane-changing completion time distribution for produced 50 000 emergency trajectories is 0.63. Among the 50 000 generated hazardous lane-changing scenarios, the collision time between the tested automated vehicle and the lane-changing background vehicle is less than 1 s in 99.54% of the scenarios. Results show that the proposed method can effectively produce hazardous lane-changing scenarios for automated vehicle testing.
- Intelligent vehicle /
- automated vehicle test /
- lane-changing scenarios /
- generation of hazardous scenarios /
- SeqGAN

HTML全文

图 1 数据采集区域

Fig. 1 Data acquisition area

下载: 全尺寸图片幻灯片

图 2 变道数据速度、加速度分析

Fig. 2 Speed and acceleration analysis of lane-changing data

下载: 全尺寸图片幻灯片

图 3 真实数据纵向速度分布

Fig. 3 Longitudinal speed distribution of real data

下载: 全尺寸图片幻灯片

图 4 真实数据速度标准差

Fig. 4 Standard deviation of speed of real data

下载: 全尺寸图片幻灯片

图 5 SeqGAN 的结构图

Fig. 5 Structure diagram of SeqGAN

下载: 全尺寸图片幻灯片

图 6 BN-AM-SeqGAN 的结构图

Fig. 6 Structure diagram of BN-AM-SeqGAN

下载: 全尺寸图片幻灯片

图 7 被测自动驾驶车辆和变道背景车行驶状态

Fig. 7 Driving status of the tested automated vehicle and lane-changing background vehicle

下载: 全尺寸图片幻灯片

图 8 变道车真实轨迹缓冲区实例

Fig. 8 Example of the real trajectory buffer of the lane-changing vehicle

下载: 全尺寸图片幻灯片

图 9 变道开始时原始速度和生成速度的状态分布

Fig. 9 State distribution of original speed and generated speed at the beginning of lane-changing

下载: 全尺寸图片幻灯片

图 10 变道结束时原始速度和生成速度的状态分布

Fig. 10 State distribution of original speed and generated speed at the end of lane-changing

下载: 全尺寸图片幻灯片

图 11 位置和速度的均方根误差

Fig. 11 Root mean square error of position and speed

下载: 全尺寸图片幻灯片

图 12 三种生成对抗网络的损失值对比

Fig. 12 The comparison of loss values of three generative adversarial networks

下载: 全尺寸图片幻灯片

图 13 危险变道测试场景

Fig. 13 Dangerous lane-changing test scenarios

下载: 全尺寸图片幻灯片

图 14 不同变道车辆速度的变道场景

Fig. 14 Lane-changing scenarios of different lane-changing speeds

下载: 全尺寸图片幻灯片

图 15 危险变道测试场景库TTC百分比

Fig. 15 TTC percentage of dangerous lane-changing scenarios library

下载: 全尺寸图片幻灯片

图 16 仿真平台搭建的虚拟变道测试场景

Fig. 16 Virtual lane-changing test scenario built by simulation platform

下载: 全尺寸图片幻灯片

表 1 真实数据的数据特征

Table 1 Data characteristics of real data

变量		平均值	方差	标准差	最小值	最大值
运动状态	速度 (m/s)	13.60	3.69	1.92	13.60	22.80
	纵向加速度(m/s²)	0.21	0.53	0.73	−5.70	5.90
	横向加速度(m/s²)	0.06	0.26	0.51	−5.01	5.93
位置分布	变道后纵向位置(m)	352.42	22 300.38	3149.33	45.56	663.51
位置分布	变道后横向位置(m)	11.32	23.03	4.80	2.49	19.04

下载: 导出CSV

表 2 实验中的参数设置

Table 2 Parameter settings in the experiment

参数含义	值
制动变减速时间段${t_2}$	0.2 s
变道背景车速度${v_1}$	生成变道轨迹的平均速度
真实轨迹序列长度$N$	20
真实轨迹总数	511
变道经过的纵向距离$d_{\rm{t}}$	生成变道轨迹的纵向距离
车长$l$	4 m
变道背景车完成变道的时间$t$	等于被测自动驾驶汽车制动时间
车辆制动最大加速度$a_{\max}$	${6 \;{\rm{m/s^2} } }$
嵌入维数	64
隐藏层数	160
预训练次数	120
生成器的初始学习率	0.04
计算奖赏的参数${\gamma}$	0.95
生成器预训练次数	150
判别器预训练次数	50

下载: 导出CSV

表 3 变道完成时间分布表

Table 3 Distribution of lane-changing completion time

变道完成时间 (s)	生成数量	BN-AM-SeqGAN 生成数据 (%)	SeqGAN 生成数据 (%)	真实数据 (%)
(1.8, 2.0]	511	53.15	53.65	53.1
(1.8, 2.0]	50 000	54.06	53.08	53.1
(1.6, 1.8]	511	32.79	33.79	31.5
(1.6, 1.8]	50 000	30.49	33.46	31.5
(1.4, 1.6]	511	10.82	9.82	12.3
(1.4, 1.6]	50 000	12.17	11.25	12.3
(1.2, 1.4]	511	2.86	2.36	2.7
(1.2, 1.4]	50 000	2.82	2.10	2.7
[1.0, 1.2]	511	0.49	0.99	0.4
[1.0, 1.2]	50 000	0.46	0.11	0.4

下载: 导出CSV

表 4 网络输出效果对比

Table 4 Comparison of network output effect

网络	符合数量	有效性
SeqGAN	44 647	71.23%
RankGAN	46 001	73.39%
BN-AM-SeqGAN	50 000	79.54%

下载: 导出CSV

参考文献(30)

[1]	李力, 王飞跃. 地面交通控制的百年回顾和未来展望. 自动化学报, 2018, 44(4): 577--583 doi: 10.16383/j.aas.2018.c170616 LI Li, WANG Fei-Yue. Ground traffic control in the past century and its future perspective. ACTA AUTOMATICA SINICA, 2018, 44(4): 577--583 doi: 10.16383/j.aas.2018.c170616
[2]	Fryman J, Matthias B. Safety of industrial robots: From conventional to collaborative applications. In: Proceedings of Conference of Robotik, 7th German Conference on Robotics. Munich, Germany: VDE, 2012.
[3]	Kalra N, Paddock S M. Driving to safety: how many miles of driving would it take to demonstrate autonomous vehicle reliability. Transportation Research Part A Policy & Practice, 2016, 94(12): 182--193
[4]	赵祥模, 承靖钧, 徐志刚, 王文威, 王润民, 王冠群, 等. 基于整车在环仿真的自动驾驶汽车室内快速测试平台. 中国公路学报, 2019, 32(6): 124--136 doi: 10.19721/j.cnki.1001-7372.2019.06.013 ZHAO Xiang-Mo, CHENG Jing-Jun, XU Zhi-Gang, WANG Wen-Wei, WANG Run-Min, WANG Guan-Qun, et al. An indoor rapid-testing platform for autonomous vehicle based on vehicle-in-the-loop simulation. China Journal of Highway and Transport, 2019, 32(6): 124--136 doi: 10.19721/j.cnki.1001-7372.2019.06.013
[5]	Riedmaier S, Ponn T, Ludwig D, Schick B, and Diermeyer F. Survey on scenario-based safety assessment of automated vehicles, IEEE access, 2020, 8: 87456–-87477 doi: 10.1109/ACCESS.2020.2993730
[6]	Sun J, Zhang H, Zhou H, Yu R, Tian Y. Scenario-Based test automation for highly automated vehicles: a review and paving the way for systematic safety assurance, IEEE Transactions on Transportation Systems, 2022, 9(23): 14088-–14103
[7]	张浩杰, 苏治宝, 杨甜甜. 基于USARSim和ROS的无人平台编队仿真系统. 自动化学报, 2021, 47(6): 1390−1400 Intelligent Zhang H, Su Z, Yang T. Design of team formation simulation system for unmanned ground vehicles based on USARSim and ROS. Acta Automatica Sinica, 2021, 47(6): 1390--1400
[8]	Wyatt S, Haering J, Feilhauer M. Current approaches in HiL-Based ADAS testing. SAE International Journal of Commercial Vehicles, 2016, 9(2): 63--69 doi: 10.4271/2016-01-8013
[9]	ISO/BS PAS 21448, Road Vehicles. Safety of the Intended Functionality, 2019.
[10]	翟强, 程洪, 黄瑞, 詹慧琴, 赵洋, 李骏. 智能汽车中人工智能算法应用及其安全综述. 电子科技大学学报, 2020, 49(04): 490--498, 510 ZHAI Qiang, CHENG Hong, HUANG Rui, ZHAN Hui-Qin, ZHAO Yang, LI Jun. Review on the application and safety of artificial intelligence algorithms in intelligent vehicles.Journal of University of Electronic Science and Technology of China, 2020, 49(04):490--498, 510
[11]	邓伟文, 李江坤, 任秉韬, 王文奇, 丁娟. 面向自动驾驶的仿真场景自动生成方法综述. 中国公路学报, 2022, 35(1): 316--333 doi: 10.3969/j.issn.1001-7372.2022.01.027 DENG Wei-Wen, LI Jiang-Kun, REN Bing-Tao, WANG Wen-Qi, DING Juan. A survey on automatic simulation scenario generation methods for autonomous driving. China Journal of Highway and Transport, 2022, 35(1): 316--333 doi: 10.3969/j.issn.1001-7372.2022.01.027
[12]	陈吉清, 舒孝雄, 兰凤崇, 王俊峰. 典型危险事故特征的自动驾驶测试场景构建. 华南理工大学学报: 自然科学版, 2021, 49(5): 1--8 CHEN Ji-Qing, SHU Xiao-Xiong, LAN Feng-Chong, WANG Jun-Feng. Construction of autonomous vehicles test scenarios with typical dangerous accident characteristics. Journal of South China University of Technology: Natural Science Edition, 2021, 49(5): 1--8
[13]	王润民, 朱宇, 赵祥模, 徐志刚, 周文帅, 刘童. 自动驾驶测试场景研究进展. 交通运输工程学报, 2021, 21(02): 21--37 doi: 10.19818/j.cnki.1671-1637.2021.02.003 WANG Run-min, ZHU Yu, ZHAO Xiang-Mo, XU Zhi-Gang, ZHOU Wen-Shuai, LIU Tong. Research progress of automatic driving test scenario. Journal of Transportation Engineering, 2021, 21(02): 21--37 doi: 10.19818/j.cnki.1671-1637.2021.02.003
[14]	朱冰, 张培兴, 赵健. 面向多维度逻辑场景的自动驾驶安全性聚类评价方法. 汽车工程, 2020, 42(11): 1458--1463, 1505 doi: 10.19562/j.chinasae.qcgc.2020.11.002 ZHU Bing, ZHANG Pei-Xing, ZHAO Jian. Autonomous driving safety cluster evaluation method for multi-dimensional logic scenarios. Automotive Engineering, 2020, 42(11): 1458--1463, 1505 doi: 10.19562/j.chinasae.qcgc.2020.11.002
[15]	Menzel T, Bagschik G, Maurer M. Scenarios for develop-ment, test and validation of automated vehicles. In: Proceedings of IEEE Intelligent Vehicles Symposium (IV). Changshu, China: IEEE, 2018. 1821−1827
[16]	Jesenski S, Stellet J E, Schiegg F, Zollner J M. Generation of scenes in intersections for the validation of highly automated driving functions. In: Proceedings of IEEE Intelligent Vehicles Symposium. Paris, France: IEEE, 2019. 502−509
[17]	Ding W, Xu M, Zhao D. Learning to collide: An adaptive safety-critical scenarios generating method. arXiv preprint arXiv: 1707.04792, 2020.
[18]	Feng S, Feng Y, Yu C, Zhang Y, Liu H X. Testing scenario library generation for connected and automated vehicles, part I: methodology. IEEE Transactions on Intelligent Transportation Systems, 2021, 22(3): 1573--1582 doi: 10.1109/TITS.2020.2972211
[19]	Feng S, Feng Y, Yu C., Zhang Y, Liu H X. (2021). Testing scenario library generation for connected and automated vehicles, part Ⅱ: case studies. IEEE Transactions on Intelligent Transportation Systems, 2021, 22(9): 5635--5647 doi: 10.1109/TITS.2020.2988309
[20]	Feng S, Feng Y, Yu C, Zhang Y, Liu H X. Intelligent driving intelligence test for autonomous vehicles with naturalistic and adversarial environment. Nature Communications, 2021, 12(1):1--14 doi: 10.1038/s41467-020-20314-w
[21]	周文帅, 朱宇, 赵祥模, 王润民, 徐志刚. 面向高速公路车辆切入场景的自动驾驶测试用例生成方法. 汽车技术, 2021, (1): 11--18 doi: 10.19620/j.cnki.1000-3703.20191450 ZHOU Wen-Shuai, ZHU Yu, ZHAO Xiang-Mo, WANG Run-Min, XU Zhi-Gang. Vehicle cut-in test case generation methods for testing of autonomous driving on highway. Automobile Technology, 2021(1): 11--18 doi: 10.19620/j.cnki.1000-3703.20191450
[22]	Sun J, Zhou H, Xi H, Zhang H, and Tian Y. Adaptive design of experiments for safety evaluation of automated vehicles. IEEE Transactions on Intelligent Transportation Systems, 2022, 9(22):14497--14508
[23]	朱宇, 赵祥模, 徐志刚, 王润民. 基于蒙特卡罗模拟的无人车高速公路变道虚拟测试场景自动生成算法. 中国公路学报, 2022, 35(3): 89--100 doi: 10.3969/j.issn.1001-7372.2022.03.009 ZHU Yu, ZHAO Xiang-Mo, XU Zhi-Gang, WANG Run-Min. Automatic generation algorithm of lane-change virtual test scenario on highways for automated vehicles using monte carlo simulation. China Journal of Highway and Transport, 2022, 35(3): 89--100 doi: 10.3969/j.issn.1001-7372.2022.03.009
[24]	Jia S, Hui F, Li S, Zhao X. LSTM-CNN for abnormal driving behavior recognition. IET Intelligent Transport Systems, 2019, 14(7): 306--312
[25]	吴斌, 朱西产, 沈剑平, 李霖. 自然驾驶工况的驾驶员紧急转向变道行为. 同济大学学报: 自然科学版, 2017(4):554--561 WU Bin, ZHU Xi-Chan, SHEN Jian-Ping, LI Lin. Analysis of driver emergency steering lane changing behavior based on naturalistic driving data. Journal of Tongji University :Natural Science, 2017(4):554--561
[26]	Yu L, Zhang W, Wang J, Yong Y. SeqGAN: Sequence generative adversarial nets with policy gradient. In: Proceedings of AAAI-17: Thirty-first AAAI Conference on Artificial Intelligence. San Francisco, California, USA: AAAI, 2017. 2852−2858
[27]	林懿伦, 戴星原, 李力, 王晓, 王飞跃. 人工智能研究的新前线: 生成式对抗网络. 自动化学报, 2018, 44(5): 775--792 LIN Yi-Lun, DAI Xing-Yuan, LI Li, WANG Xiao, WANG Fei-Yue. The new frontier of AI research: generative adversarial networks. ACTA AUTOMATICA SINICA, 2018, 44(5): 775--792
[28]	王坤峰, 左旺孟, 谭营, 秦涛, 李力, 王飞跃.生成式对抗网络:从生成数据到创造智能. 自动化学报, 2018, 44(5): 769--774 doi: 10.16383/j.aas.2018.y000001 WANG Kun-Feng, ZUO Wang-Meng, TAN Ying, QIN Tao, LI Li, WANG Fei-Yue. Generative adversarial networks: from generating data to creating intelligence. ACTA AUTOMATICA SINICA, 2018, 44(5): 769--774 doi: 10.16383/j.aas.2018.y000001
[29]	Zhao X, Jing S, Hu F, Liu R, Khattak J A. DSRC-based rear-end collision warning system - an error- component safety distance model and field test. Transportation research part C: Emerging Technologies, 2019, 107: 92--104 doi: 10.1016/j.trc.2019.08.002
[30]	交通部公路司中国工程建设标准化协会公路工程委员会. 公路工程技术标准. 人民交通出版社, 2004. Highway Department of Ministry of Communications, Highway Engineering Committee of China Engineering Construction Standardization Association. Technical Standard of Highway Engineering. People＇s Communications Press, 2004.