雷达与视觉传感器融合的高速公路全域车辆轨迹与交通参数检测方法

戴喆; 吴宇轩; 董是; 王建伟; 袁长伟; 左琛

doi:10.19818/j.cnki.1671-1637.2025.01.014

雷达与视觉传感器融合的高速公路全域车辆轨迹与交通参数检测方法

doi: 10.19818/j.cnki.1671-1637.2025.01.014

戴喆^{1, 2},
吴宇轩^{1, 2},
董是^{1, 2, ,},
王建伟^{1, 2},
袁长伟^{1, 2},
左琛^{1, 2}

1.
长安大学运输工程学院，陕西西安 710064
2.
长安大学道路基础设施数字化教育部工程研究中心，陕西西安 710064

基金项目:

国家自然科学基金项目 52402399

陕西省自然科学基础研究计划项目 2024JC-YBQN-0369

中国博士后科学基金项目 2022M710482

浙江省交通运输厅科技计划项目 2023016

详细信息

作者简介:
戴喆(1993-), 男, 陕西西安人, 长安大学讲师, 工学博士, 从事智能交通系统研究

通讯作者:
董是(1989-), 男, 辽宁辽阳人, 长安大学副教授, 工学博士

中图分类号: U491.1
计量
- 文章访问数: 27
- HTML全文浏览量: 30
- PDF下载量: 12
- 被引次数: 0
出版历程
- 收稿日期: 2023-11-06
- 刊出日期: 2025-02-25

Global vehicle trajectories and traffic parameters detecting method in expressway based on radar and vision sensor fusion

DAI Zhe^{1, 2},
WU Yu-xuan^{1, 2},
DONG Shi^{1, 2
, ,},
WANG Jian-wei^{1, 2},
YUAN Chang-wei^{1, 2},
ZUO Chen^{1, 2}

1.
School of Transportation Engineering, Chang'an University, Xi'an 710064, Shaanxi, China
2.
Engineering Research Center of Highway Infrastructure Digitalization of Ministry of Education, Chang'an University, Xi'an 710064, Shaanxi, China

Funds:

National Natural Science Foundation of China 52402399

Natural Science Basic Research Project of Shaanxi Province 2024JC-YBQN-0369

China Postdoctoral Science Foundation 2022M710482

Science and Technology Project of Department of Transportation of Zhejiang Province 2023016

More Information

Corresponding author: DONG Shi(1989-), male, associate professor, PhD, dongshi@chd.edu.cn

Article Text (Baidu Translation)

摘要

摘要: 为满足智慧高速公路在复杂交通环境下对交通参数的大范围检测需求，提出了一种毫米波雷达与视觉传感器数据融合的全域车辆轨迹与交通参数检测方法；利用部署于不同路侧立杆的毫米波雷达与视觉传感器采集原始数据，通过对多源检测目标数据进行时空同步、关联、融合及多目标追踪，设计了局部场景车辆轨迹融合检测算法；通过重构车辆运动时空信息，对多个连续不同场景的车辆轨迹进行合并，设计了连续多场景联动的全域车辆轨迹检测算法；根据全域车辆轨迹中提取的位置、速度等微观运动信息，设计了基于全域车辆轨迹的交通参数检测方法；在智慧高速公路试点建设路段进行试验数据采集与人工标注，对所提出方法进行验证。研究结果表明：在目标检测任务与轨迹追踪任务中，各局部场景与连续多场景的目标检测精度整体大于90%，追踪轨迹位置与车辆实际位置的偏差均值不超过0.2 m；在交通参数检测任务中，车辆在观测区域内检测速度与实际速度的平均绝对误差加权均值为3.41 km·h^-1，平均绝对百分比误差加权均值为5.00%；区间平均速度、交通流量及交通密度等交通参数的检测精度可达车道级，检测结果与高速公路出口匝道及分流区的真实交通现象相一致。
- 智能交通 /
- 交通检测 /
- 多源数据融合 /
- 多目标追踪 /
- 时空信息重构 /
- 轨迹合并
Abstract: To meet the demand of smart expressways for a wide-range detection of traffic parameters in complex traffic environments, a global vehicle trajectories and traffic parameters detecting method was proposed based on the data fusion of millimeter-wave radar and vision sensor. Raw data was collected by millimeter-wave radars and vision sensors deployed on different roadside poles. Through spatio-temporal synchronization, association, fusion, and multi-object tracking of multi-source detection target data, an algorithm was designed for detecting vehicle trajectories in the local scene. An algorithm for detecting global vehicle trajectories in continuous scenes was designed by reconstructing vehicle spatio-temporal information and merging vehicle trajectories in continuous traffic scenes. A method for detecting traffic parameters based on global vehicle trajectories was designed through position, speed, and other microscopic motion information extracted from global vehicle trajectories. Experimental data was collected and manually labeled in a pilot section of the smart expressways and was used to validate the proposed method. Research results show that in vehicle detection task and trajectory tracking task, the overall vehicle detection accuracy of each local scene and multiple continuous traffic scenes is greater than 90%, and the deviation between the tracking trajectory position and the actual position of the vehicle is less than 0.2 m. In the traffic parameters detection task, the weighted average of mean absolute error between the detected vehicle's speed and the actual speed in the observation area is 3.41 km·h^-1, and the weighted average of mean absolute percentage error is 5.00%. The detection accuracies of traffic parameters such as space mean speed, traffic volume and traffic density can reach lane level, and the detection results are consistent with the real traffic phenomena in the expressway exit ramp and diverging area.
- intelligent transportation /
- traffic detection /
- multi-source data fusion /
- multi-object tracking /
- spatio-temporal information reconstruction /
- trajectory merging

HTML全文

图 1 雷达和视觉传感器融合的高速公路全域车辆轨迹与交通参数检测方法流程

Figure 1. Process of global vehicle trajectories and traffic parameters detecting method in expressway based on radar and vision sensor fusion

下载: 全尺寸图片幻灯片

图 2 毫米波雷达与视觉传感器工作姿态

Figure 2. Working posture of millimeter-wave radar and vision sensor

下载: 全尺寸图片幻灯片

图 3 毫米波雷达与视觉传感器数据时间同步流程

Figure 3. Time synchronization process of data from millimeter-wave radar and vision sensor

下载: 全尺寸图片幻灯片

图 4 多源检测车辆目标关联状态

Figure 4. Association status of multi-source detection of vehicle targets

下载: 全尺寸图片幻灯片

图 5 局部场景车辆轨迹时空分布

Figure 5. Spatio-temporal distributions of vehicle trajectories in each local scene

下载: 全尺寸图片幻灯片

图 6 连续多场景联动的全域车辆轨迹检测算法流程

Figure 6. Process of global vehicle trajectories detection algorithm in continuous traffic scenes

下载: 全尺寸图片幻灯片

图 7 连续多场景联动的全域车辆轨迹时空分布

Figure 7. Spatio-temporal distribution of global vehicle trajectories in continuous traffic scenes

下载: 全尺寸图片幻灯片

图 8 数据采集区域

Figure 8. Data collection areas

下载: 全尺寸图片幻灯片

图 9 不同检测方法在各观测点追踪车辆轨迹的可视化示例

Figure 9. Visualization examples of vehicle trajectories tracked by different detecting methods at each observation site

下载: 全尺寸图片幻灯片

图 10 检测速度与实际速度对比结果

Figure 10. Comparison results of detected speeds and actual speeds

下载: 全尺寸图片幻灯片

表 1 各观测点视觉传感器参数标定结果

Table 1. Calibration results of vision sensor parameters at each observation site

观测点	1	2	3	4
θ/rad	-0.140 23	-0.115 54	-0.129 50	-0.125 06
φ/rad	0.101 05	0.066 32	0.070 69	0.085 63
f/mm	6 500.05	8 500.00	6 455.00	6 756.62
H/mm	10 709.45	6 000.00	7 000.00	7 000.00

下载: 导出CSV

表 2 基于全域车辆轨迹提取的微观车辆运动参数

Table 2. Microscopic vehicle motion parameters extracted from global vehicle trajectories

数据字段名	数据描述
track_id	车辆编号
class	车辆类别
time_stamp	检测到该车辆时的Unix时间戳/ms
xPosition	车辆在UTM坐标系下的东向位置坐标/m
yPosition	车辆在UTM坐标系下的北向位置坐标/m
xSpeed	车辆在UTM坐标系下的东向速度/(m·s^-1)
ySpeed	车辆在UTM坐标系下的北向速度/(m·s^-1)
driveAngle	车辆在UTM坐标系下的行驶方向角/(°)
sub_node_id	检测到该车辆的观测点编号

下载: 导出CSV

表 3 数据采集时间内不同检测方法的准确率

Table 3. Accuracies for different detecting methods during data collection time %

观测点	1		2		3		4
观测点	近端	远端	近端	远端	近端	远端	近端	远端
E_{Acc, 1}	98.61	80.79	98.43	78.65	98.14	77.39	98.52	78.13
E_{Acc, 2}	84.26	98.74	83.31	96.92	82.46	96.55	83.99	96.61
E_{Acc, 3}	97.52	97.13	95.01	94.42	94.13	92.38	95.38	93.86

下载: 导出CSV

表 4 各观测点车辆检测目标试验结果

Table 4. Experimental results of vehicle detection task at each observation site

观测点	1	2	3	4
E_GT	7 293	5 782	10 228	5 251
E_TP	7 094	5 467	9 535	4 942
E_FP1	89	153	391	161
E_FN1	110	162	302	148
E_Acc/%	97.27	94.55	93.22	94.12
E_Pre/%	98.76	97.12	96.06	96.84
E_Rc/%	98.47	97.27	96.93	97.09

下载: 导出CSV

表 5 车辆运动轨迹追踪任务试验结果

Table 5. Experimental results of vehicle trajectories tracking task

观测点	1	2	3	4	全域
E_GT	7 293	5 782	10 228	5 251	28 554
E_TP	7 094	5 467	9 535	4 942	27 038
E_FP2	89	153	391	161	794
E_FN2	110	162	302	148	722
E_IDSW	5	3	1	2	11
E_MOTA/%	97.20	94.50	93.21	94.08	94.65
E_MOTP/%	0.20	0.16	0.21	0.16	0.18

下载: 导出CSV

表 6 各观测点车辆速度E_MAE及E_MAPE

Table 6. MAE and MAPE in vehicle speed detection

观测点	1	2	3	4	全域
E_MAE/(km·h^-1)	2.72	4.31	3.04	3.73	3.41
E_MAPE/%	4.32	6.03	4.43	5.29	5.00

下载: 导出CSV

表 7 数据采集时间内各车道不同类型车辆的区间平均速度

Table 7. Space mean speeds for different types of vehicles in each lane during data collection time km·h^-1

车辆类型	车道1		车道2		车道3		车道4
车辆类型	本文算法	人工测得	本文算法	人工测得	本文算法	人工测得	本文算法	人工测得
所有车辆	57.13	59.08	69.95	71.56	73.73	74.99	82.60	84.34
大型车辆	56.94	57.92	69.55	71.15	72.75	73.84	81.28	83.14
小型车辆	57.28	59.16	69.99	71.61	73.74	75.06	83.04	84.48

下载: 导出CSV

表 8 数据采集时间内各车道不同类型车辆的交通密度

Table 8. Traffic densities for different types of vehicles in each lane during data collection time veh·km^-1

车辆类型	车道1		车道2		车道3		车道4
车辆类型	本文算法	人工测得	本文算法	人工测得	本文算法	人工测得	本文算法	人工测得
所有车辆	7.08	7.20	6.74	6.83	5.87	6.23	3.37	3.56
大型车辆	0.51	0.53	0.22	0.25	0.05	0.06	0.38	0.39
小型车辆	6.55	6.67	6.52	6.58	5.82	6.17	2.99	3.17

下载: 导出CSV

表 9 数据采集时间内各车道不同类型车辆的交通流量

Table 9. Traffic volumes for different types of vehicles in each lane during data collection time veh·h^-1

车辆类型	车道1		车道2		车道3		车道4
车辆类型	本文算法	人工测得	本文算法	人工测得	本文算法	人工测得	本文算法	人工测得
所有车辆	404.5	425.3	471.5	488.8	432.8	467.2	278.4	300.3
大型车辆	29.3	30.7	15.2	17.7	3.5	4.1	30.3	32.5
小型车辆	375.2	394.6	456.3	471.1	429.3	463.1	248.1	267.8

下载: 导出CSV

参考文献(30)

[1]	ZHOU Wei, YU Yun-hong, ZHAN Yun-fei, et al. A vision-based abnormal trajectory detection framework for online traffic incident alert on freeways[J]. Neural Computing and Applications, 2022, 34(17): 14945-14958. doi: 10.1007/s00521-022-07335-w
[2]	ZHANG Yun-chao, CHEN Yan-yan, GU Xin, et al. A proactive crash risk prediction framework for lane-changing behavior incorporating individual driving styles[J]. Accident Analysis and Prevention, 2023, 188: 107072-107089. doi: 10.1016/j.aap.2023.107072
[3]	吕能超, 彭凌枫, 吴超仲, 等. 基于视频轨迹参数的边缘率减速标线驾驶行为效果评价方法[J]. 安全与环境学报, 2021, 21(2): 461-469. LYU Neng-chao, PENG Ling-feng, WU Chao-zhong, et al. Evaluative judgment of the driving behaviors of the edge rate deceleration bars based on the video trajectory parameters[J]. Journal of Safety and Environment, 2021, 21(2): 461-469.
[4]	MA Yue-xin, ZHU Xin-ge, ZHANG Si-bo, et al. TrafficPredict: Trajectory prediction for heterogeneous traffic agents[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2019, 33(1): 6120-6127. doi: 10.1609/aaai.v33i01.33016120
[5]	LIANG Hao-xiang, SONG Huan-sheng, YUN Xu, et al. Traffic incident detection based on a global trajectory spatiotemporal map[J]. Complex and Intelligent Systems, 2022, 8(2): 1389-1408. doi: 10.1007/s40747-021-00602-8
[6]	ZHANG T Y, JIN P J. A longitudinal scanline based vehicle trajectory reconstruction method for high-angle traffic video[J]. Transportation Research Part C: Emerging Technologies, 2019, 103: 104-128. doi: 10.1016/j.trc.2019.03.015
[7]	ZHANG Jia-xing, WEN Xiao, COIFMAN B, et al. Vehicle tracking and speed estimation from roadside lidar[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2020, 13: 5597-5608. doi: 10.1109/JSTARS.2020.3024921
[8]	王俊骅, 宋昊, 景强, 等. 基于毫米波雷达组群的全域车辆轨迹检测技术方法[J]. 中国公路学报, 2022, 35(12): 181-192. WANG Jun-hua, SONG Hao, JING Qiang, et al. Road-range tracking of vehicle trajectories based on millimeter-wave radar[J]. China Journal of Highway and Transport, 2022, 35(12): 181-192.
[9]	ZHOU Yong, DONG Yan-yan, HOU Fu-jin, et al. Review on millimeter-wave radar and camera fusion technology[J]. Sustainability, 2022, 14(9): 5114.
[10]	邓淇天, 李旭. 基于多特征融合的车辆检测算法[J]. 传感器与微系统, 2020, 39(6): 131-134. DENG Qi-tian, LI Xu. Vehicle detection algorithm based on multi-feature fusion[J]. Transducer and Microsystem Technologies, 2020, 39(6): 131-134.
[11]	甘耀东, 郑玲, 张志达, 等. 融合毫米波雷达与深度视觉的多目标检测与跟踪[J]. 汽车工程, 2021, 43(7): 1022-1029, 1056. GAN Yao-dong, ZHENG Ling, ZHANG Zhi-da, et al. Multi-target detection and tracking with fusion of millimeter-wave radar and deep vision[J]. Automotive Engineering, 2021, 43(7): 1022-1029, 1056.
[12]	贾鹏飞, 刘全周, 彭凯, 等. 多传感器信息融合的前方车辆检测[J]. 红外与激光工程, 2022, 51(6): 20210446. JIA Peng-fei, LIU Quan-zhou, PENG Kai, et al. Front vehicle detection based on multi-sensor information fusion[J]. Infrared and Laser Engineering, 2022, 51(6): 20210446.
[13]	武一民, 郑凯元, 高博麟, 等. 基于自适应扩展Kalman滤波的路侧多传感器融合[J]. 汽车安全与节能学报, 2021, 12(4): 522-527. WU Yi-min, ZHENG Kai-yuan, GAO Bo-lin, et al. Roadside multi-sensor fusion based on adaptive extended Kalman filter[J]. Journal of Automotive Safety and Energy, 2021, 12(4): 522-527.
[14]	WANG Shu-jian, PI Ren-dong, LI Jian, et al. Object tracking based on the fusion of roadside LiDAR and camera data[J]. IEEE Transactions on Instrumentation and Measurement, 2022, 71: 7006814.
[15]	金沙沙, 龙伟, 胡灵犀, 等. 多目标检测与跟踪算法在智能交通监控系统中的研究进展[J]. 控制与决策, 2023, 38(4): 890-901. JIN Sha-sha, LONG Wei, HU Ling-xi, et al. Research progress of detection and multi-object tracking algorithm in intelligent traffic monitoring system[J]. Control and Decision, 2023, 38(4): 890-901.
[16]	刘超, 罗如意, 刘春青, 等. 基于路侧多机视频目标关联与轨迹拼接的车辆连续轨迹构建方法[J]. 交通信息与安全, 2023, 41(3): 80-91. LIU Chao, LUO Ru-yi, LIU Chun-qing, et al. A method for developing continuous vehicle trajectories through target association and trajectory splicing based on video data from multiple roadside cameras[J]. Journal of Transport Information and Safety, 2023, 41(3): 80-91.
[17]	耿华, 张涛, 连小珉, 等. 北斗卫星与车载传感器融合的车辆定位[J]. 汽车工程, 2007, 29(11): 929-933. GENG Hua, ZHANG Tao, LIAN Xiao-min, et al. Vehicle positioning by Beidou satellite fusing with on-board sensors[J]. Automotive Engineering, 2007, 29(11): 929-933.
[18]	ELAZAB M, NOURELDIN A, HASSANEIN H S. Integrated cooperative localization for vehicular networks with partial GPS access in urban canyons[J]. Vehicular Communications, 2017, 9: 242-253.
[19]	PEREA-STROM D, MORELL A, TOLEDO J, et al. GNSS integration in the localization system of an autonomous vehicle based on particle weighting[J]. IEEE Sensors Journal, 2020, 20(6): 3314-3323.
[20]	KIM K, IM J, JEE G. Tunnel facility based vehicle localization in highway tunnel using 3D LIDAR[J]. IEEE Transactions on Intelligent Transportation Systems, 2022, 23(10): 17575-17583.
[21]	CORNICK M, KOECHLING J, STANLEY B, et al. Localizing ground penetrating radar: a step toward robust autonomous ground vehicle localization[J]. Journal of Field Robotics, 2016, 33(1): 82-102.
[22]	邓泽武, 胡钊政, 周哲, 等. 融合激光雷达与双层地图模型的智能车定位[J]. 汽车工程, 2022, 44(7): 1018-1026. DENG Ze-wu, HU Zhao-zheng, ZHOU Zhe, et al. Intelligent vehicle positioning by fusing LiDAR and double-layer map model[J]. Automotive Engineering, 2022, 44(7): 1018-1026.
[23]	ADEGOKE E I, ZIDANE J, KAMPERT E, et al. Infrastructure Wi-Fi for connected autonomous vehicle positioning: a review of the state-of-the-art[J]. Vehicular Communications, 2019, 20: 100185.
[24]	王伟, 张朝阳, 唐心瑶, 等. 道路场景下相机自动标定及优化算法[J]. 计算机辅助设计与图形学学报, 2019, 31(11): 1955-1962. WANG Wei, ZHANG Chao-yang, TANG Xin-yao, et al. Automatic self-calibration and optimization algorithm of traffic camera in road scene[J]. Journal of Computer-Aided Design and Computer Graphics, 2019, 31(11): 1955-1962.
[25]	辜志强, 吉鑫钰, 褚端峰, 等. 基于全局位置精度损失最小化的路侧多传感器目标关联匹配方法[J]. 中国公路学报, 2022, 35(3): 286-294. GU Zhi-qiang, JI Xin-yu, CHU Duan-feng, et al. A roadside multi-sensor target association matching method based on minimization of global position precision loss[J]. China Journal of Highway and Transport, 2022, 35(3): 286-294.
[26]	KE Guo-lin, MENG Qi, FINLEY T, et al. LightGBM: A highly efficient gradient boosting decision tree[J]. Advances in neural information processing systems, 2017, 2017: 3147-3155.
[27]	WOJKE N, BEWLEY A, PAULUS D. Simple online and realtime tracking with a deep association metric[C]//IEEE. 24th IEEE International Conference on Image Processing (ICIP). New York: IEEE, 2017: 3645-3649.
[28]	CHAN Si-xian, QIU Chen-hao, WU Di-juan, et al. Fusion detection and ReID embedding with hybrid attention for multi-object tracking[J]. Neurocomputing, 2024, 575: 127328.
[29]	KANHERE N K, BIRCHFIELD S T. A taxonomy and analysis of camera calibration methods for traffic monitoring applications[J]. IEEE Transactions on Intelligent Transportation Systems, 2010, 11(2): 441-452.
[30]	LUO Wen-han, XING Jun-liang, MILAN A, et al. Multiple object tracking: a literature review[J]. Artificial intelligence, 2021, 293: 103448-103471.