融合高精度地图的多激光雷达路侧智能感知方法

胡钊政; 陈琪莉; 孟杰; 胡华桦; 张佳楠

doi:10.3963/j.jssn.1674-4861.2024.03.005

融合高精度地图的多激光雷达路侧智能感知方法

doi: 10.3963/j.jssn.1674-4861.2024.03.005

胡钊政^{1, 2,},
陈琪莉^{1, 2},
孟杰^{1, 3, ,},
胡华桦^{1, 2},
张佳楠^{1, 2}

1.
武汉理工大学智能交通系统研究中心武汉 430063
2.
武汉理工大学重庆研究院重庆 401120
3.
武汉理工大学交通物联网技术湖北省重点实验室武汉 430070

基金项目:

国家重点研发计划项目 2021YFB2501104

湖北省重点研发计划项目 2022BAA082

重庆市科技创新重大研发项目 CSTB2020TIAD-STX0003

详细信息

作者简介:
胡钊政（1979—），博士，教授. 研究方向：车路协同、智能网联汽车等. E-mail：zzhu@whut.edu.cn

通讯作者:
孟杰（1993—），助理研究员. 研究方向：自动驾驶、移动机器人等. E-mail：mengjie09@whut.edu.cn

中图分类号: U495+TN958.98
计量
- 文章访问数: 160
- HTML全文浏览量: 99
- PDF下载量: 9
- 被引次数: 0
出版历程
- 收稿日期: 2023-12-19
- 网络出版日期: 2024-10-21

Multi-LiDAR Roadside Intelligent Perception Method Fusing High-Definition Map

HU Zhaozheng^{1, 2
,},
CHEN Qili^{1, 2},
MENG Jie^{1, 3
, ,},
HU Huahua^{1, 2},
ZHANG Jianan^{1, 2}

1.
Intelligent Transportation Systems Research Center, Wuhan University of Technology, Wuhan 430063, China
2.
Chongqing Research Institute, Wuhan University of Technology, Chongqing 401120, China
3.
Hubei Key Laboratory of Transportation Internet of things, Wuhan University of Technology, Wuhan 430070, China

摘要

摘要: 在车路协同路侧感知研究中，由于点云数据量庞大且存在着不可避免的目标遮挡情况，导致检测效率低、目标轨迹不稳定和跟踪精度低的问题。针对上述问题，提出了1种融合高精度地图的多激光雷达路侧智能感知方法，通过融合高精度地图信息，提高感知结果的准确性和可靠性。该方法分为3个部分：①通过多激光雷达的标定结果，利用高精度地图完成三维点云区域中感兴趣区域的提取，从而有效减少待处理点云的数量，提升计算效率；②基于极化图高斯混合背景模型的背景建模方法，利用极化图完成运动目标快速检测，避免大规模激光点云的直接处理，有效提升检测效率；③利用车辆航向与车道线方向一致性约束，将高精度地图中的车道方向转化为卡尔曼滤波框架下的车辆状态线性约束，改善车辆检测与轨迹跟踪的性能。实验中，分别在仿真交叉路口与实车实验道路双T形路口对算法与模型进行测试验证。相比于其他方法，所提出的方法数据量减少了55%，目标检测准确率提高了12%，耗时减少了56%，目标跟踪的误差极值、误差均值以及均方根误差均有所降低。实验结果表明：所提的方法能有效融合高精度地图信息，在大范围道路场景下实现对道路运动目标的快速检测与稳定跟踪。
- 智能交通 /
- 路侧感知 /
- 目标跟踪 /
- 卡尔曼滤波 /
- 高精度地图 /
- 极化图
Abstract: In the research of vehicle-road collaborative roadside perception, challenges such as low detection efficiency, unstable target trajectories, and inaccurate tracking arise due to the sheer volume of point cloud data and the inevitable obstruction of targets. To tackle these issues, a method of intelligent roadside perception utilizing multi-LiDAR fused with High-Definition (HD) maps is proposed. The goal is to enhance the accuracy and reliability of perception outcomes by incorporating detailed map information. Leveraging the calibration results of multi-LiDAR, the extraction of the region of interest (ROI) within the three-dimensional point cloud is achieved through HD maps, effectively reducing the quantity of point clouds for processing and enhancing computational efficiency. Employing the polar-image Gaussian mixture model (P-GMM) for background modeling, moving targets are swiftly identified using polar-images to circumvent direct processing of extensive LiDAR point clouds, thereby boosting detection efficiency. By enforcing the alignment between vehicle heading and lane line direction, the lane orientation in the HD map is translated into a linear constraint of vehicle state within the Kalman filter framework, thereby enhancing the efficacy of vehicle detection and trajectory tracking. Experimental validation is conducted using simulated crossroads and real-world roads with double T-shaped intersections. Compared to other methods, the method proposed yielded a 55% reduction in data volume, a 12% increase in target detection accuracy, and a 56% decrease in processing time. The improvements in extreme error, mean error, and root mean square error are also achieved in target tracking. The experimental results show that the method proposed can fuse HD map information effectively, achieving rapid detection and tracking of road-moving targets in a wide range of road scenarios.
- intelligent transportation /
- roadside perception /
- target tracking /
- Kalman filter /
- High-Definition map /
- polar-image

HTML全文

图 1 算法流程图

Figure 1. Algorithm flowchart

下载: 全尺寸图片幻灯片

图 2 利用高精度地图完成ROI区域提取

Figure 2. ROI area extraction with HD map

下载: 全尺寸图片幻灯片

图 3 车辆航向角与车道线方向角度

Figure 3. Vehicle heading angle and lane marking angle

下载: 全尺寸图片幻灯片

图 4 仿真实验场景及对应的高精度地图

Figure 4. Schematic diagram of simulation experiment scene and HD map

下载: 全尺寸图片幻灯片

图 5 仿真平台的多激光雷达标定结果

Figure 5. Multi-LiDAR calibration results of simulation platform

下载: 全尺寸图片幻灯片

图 6 基于ROI提取的目标检测过程

Figure 6. Target detection process based on ROI extraction

下载: 全尺寸图片幻灯片

图 7 基于仿真平台的多激光雷达目标检测效果展示

Figure 7. Mmulti-LiDAR target detection effect based on simulation platform

下载: 全尺寸图片幻灯片

图 8 基于仿真平台的ID为27的车辆跟踪轨迹

Figure 8. Vehicle tracking trajectory with ID 27 based on simulation platform

下载: 全尺寸图片幻灯片

图 9 基于仿真平台的车辆跟踪结果算法对比

Figure 9. Comparison of vehicle tracking results algorithms based on simulation platform

下载: 全尺寸图片幻灯片

图 10 实车实验场景示意图及高精度地图

Figure 10. Schematic diagram and HD map of the real vehicle experiment scene

下载: 全尺寸图片幻灯片

图 11 实车实验场景下多激光雷达现场架构

Figure 11. Multi-LiDAR field architecture in real vehicle experiment scene

下载: 全尺寸图片幻灯片

图 12 实车实验场景下多激光雷达标定及点云拼接结果

Figure 12. Multi-LiDAR calibration and point cloud combination results in real vehicle experiment scene

下载: 全尺寸图片幻灯片

图 13 实车实验场景目标检测效果

Figure 13. Target detection effect of the real vehicle experiment scene

下载: 全尺寸图片幻灯片

图 14 基于实车实验场景的ID为2的车辆跟踪轨迹

Figure 14. Vehicle tracking trajectory with ID 2 based on real vehicle experiment scene

下载: 全尺寸图片幻灯片

图 15 基于实车实验场景的车辆跟踪结果算法对比

Figure 15. Comparison of vehicle tracking results algorithms based on real vehicle experiment scene

下载: 全尺寸图片幻灯片

表 1 基于仿真平台的目标检测算法对比

Table 1. Comparison of target detection algorithms based on simulation platform

数据帧	方法	目标数量	错误检测	准确率/%	耗时/ms
27	基于整个场景的目标检测	6	1	83.33	127
27	本文算法	6	0	100.00	53
331	基于整个场景的目标检测	16	3	81.25	158
331	本文算法	16	1	93.75	68
756	基于整个场景的目标检测	21	3	85.71	174
756	本文算法	21	1	95.24	79
合计	基于整个场景的目标检测	43	7	83.72	459
合计	本文算法	43	2	95.35	200

下载: 导出CSV

表 2 仿真环境下不同算法车辆跟踪误差对比分析

Table 2. Comparative analysis of vehicle tracking errors in simulation environments

方法	方向	误差极值/cm	误差均值/cm	均方根误差/cm
文献[8]	x	19.3	6.4	10.9
文献[8]	y	17.5	7.0	10.2
文献[9]	x	24.6	8.9	12.6
文献[9]	y	29.0	7.6	12.8
文献[10]	x	18.9	5.7	9.2
文献[10]	y	17.2	6.8	9.8
本文算法	x	16.6	4.1	7.0
本文算法	y	13.4	4.3	6.4

下载: 导出CSV

表 3 基于实车实验场景的目标检测算法对比

Table 3. Comparison of target detection algorithms based on real vehicle experiment scene

数据帧	方法	目标数量	错误检测	准确率/%	耗时/ms
12	基于整个场景的目标检测	3	1	66.67	94
12	本文算法	3	0	100.00	55

下载: 导出CSV

表 4 基于实车实验场景的车辆跟踪误差对比分析

Table 4. Comparative analysis of vehicle tracking errors based on real vehicle experiment scene

方法	方向	误差极值/cm	误差均值/cm	均方根误差/cm
文献[8]	x	26.1	9.4	14.8
文献[8]	y	25.9	9.8	14.4
文献[9]	x	34.7	13.3	18.5
文献[9]	y	36.2	12.5	17.9
文献[10]	x	23.6	8.3	12.7
文献[10]	y	24.8	8.1	12.2
本文算法	x	18.6	6.7	10.7
本文算法	y	20.4	7.2	10.3

下载: 导出CSV

参考文献(23)

[1]	张亚勤, 李震宇, 尚国斌, 等. 面向自动驾驶的车路云一体化框架[J]. 汽车安全与节能学报, 2023, 14(3): 249-273. doi: 10.3969/j.issn.1674-8484.2023.03.001 ZHANG Y Q, LI Z Y, SHANG G B, et al. A unified framework for vehicle-infrastructure-cloud autonomous driving[J]. Journal of Automotive Safety and Energy Saving, 2023, 14(3): 249-273. (in Chinese) doi: 10.3969/j.issn.1674-8484.2023.03.001
[2]	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 Z Q, JI X Y, CHU D F, 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. (in Chinese)
[3]	任柯燕, 谷美颖, 袁正谦, 等. 自动驾驶3D目标检测研究综述[J]. 控制与决策, 2023, 38(4): 865-889. REN K Y, GU M Y, YUAN Z Q, et al. 3D object detection algorithms in autonomous driving: a review[J]. Control and Decision, 2023, 38(4): 865-889. (in Chinese)
[4]	BELTRAN J, GUNIDEL C, MORENO F M, et al. BirdNet: a 3D object detection framework from LiDAR information[C]. 21^st International Conference on Intelligent Transportation Systems(ITSC). Maui, HI: IEEE, 2018.
[5]	孙挺, 齐迎春, 耿国华. 基于帧间差分和背景差分的运动目标检测算法[J]. 吉林大学学报(工学版), 2016, 46(4): 1325-1329. SUN T, QI Y C, GENG G H. Moving object detection algorithm based on frame difference and background subtraction[J]. Journal of Jilin University(Engineering and Technology Edition), 2016, 46(4): 1325-1329. (in Chinese)
[6]	ZHANG Z, ZHENG J, XU H, et al. Automatic background construction and object detection based on roadside LiDAR[J]. IEEE Transactions on Intelligent Transportation Systems, 2020, 21(10): 4086-4097. doi: 10.1109/TITS.2019.2936498
[7]	李杰, 张洛维, 王晓燕, 等. 基于视锥距离和自适应权重卡尔曼滤波的多传感器融合算法[J]. 中国公路学报, 2024, 37: 194-203. LI J, ZHANG L W, WANG X Y, et al. A multi-sensor fusion algorithm based on view-cone distance and adaptive weighted Kalman filter[J]. China Journal of Highway and Transport, 2024, 37: 194-203. (in Chinese)
[8]	徐国艳, 牛欢, 郭宸阳, 等. 基于三维激光点云的目标识别与跟踪研究[J]. 汽车工程, 2020, 42(1): 38-46. XU G Y, NIU H, GUO C Y, et al. Research on target recognition and tracking based on 3D laser point cloud[J]. Automotive Engineering, 2020, 42(1): 38-46. (in Chinese)
[9]	ZHAO J, XU H, LIU H, et al. Detection and tracking of pedestrians and vehicles using roadside LiDAR sensors[J]. Transportation Research Part C: Emerging Technologies, 2019, 100: 68-87. doi: 10.1016/j.trc.2019.01.007
[10]	LIN C, WANG Y, GONG B, et al. Vehicle detection and tracking using low-channel roadside LiDAR[J]. Measurement, 2023, 218: 113159.
[11]	YANG B, LIANG M, URTASUN R. HDNET: exploiting HD maps for 3D object detection[C]. 2^nd Conference on Robot Learning, Zürich, Switzerland: PMLR, : 2018.
[12]	杨振凯, 华一新, 訾璐, 等. 浅析高精度地图发展现状及关键技术[J]. 测绘通报, 2021(6): 54-60. YANG Z K, HUA Y X, ZI L, et al. Analysis of the development status and key technologies of high-precision map[J]. Bulletin of Surveying and Mapping, 2021(6): 54-60. (in Chinese)
[13]	SEIF H G, HU X. Autonomous driving in the ICity—HD maps as a key challenge of the automotive industry[J]. Engineering, 2016, 2(2): 159-162.
[14]	MA W C, URTASUN R, TARTAVULL I, et al. Exploiting sparse semantic HD maps for self-driving vehicle localization[C]. IEEE/RSJ International Conference on Intelligent Robots and Systems(IROS). Macau, China: IEEE, 2019.
[15]	CAI H, HU Z, HUANG G, et al. Integration of GPS, monocular vision, and high definition(HD)map for accurate vehicle localization[J]. Sensors, 2018, 18(10): 3270.
[16]	GHALLABI F, NASHASHIBI F, EI-HAJ-SHHADE G, et al. LiDAR-based lane marking detection for vehicle positioning in an HD map[C]. 21^st International Conference on Intelligent Transportation Systems(ITSC). Maui, HI: IEEE, 2018.
[17]	胡钊政, 孙勋培, 张佳楠, 等. 基于时空图模型的车-路-图协同定位方法[J]. 吉林大学学报(工学版), 2024, 54(5): 1246-1257. HU Z Z, SUN X P, ZHANG J N, et al. Vehicle-infrastructure-map cooperative localization method based on spatial-temporal graph model[J]. Journal of Jilin University(Engineering and Technology Edition), 2024, 54(5): 1246-1257. (in Chinese)
[18]	BAUER S, ALKHORSHID Y, WANIELIK G. Using high-definition maps for precise urban vehicle localization[C]. 19^th International Conference on Intelligent Transportation Systems (ITSC). Rio de Janeiro, Brazil: IEEE, 2016.
[19]	JIANG K, YAND D, LIU C, et al. A flexible multi-layer map model designed for lane-level route planning in autonomous vehicles[J]. Engineering, 2019, 5(2): 305-318.
[20]	王丞, 田暄, 郭瑞, 等. 自适应Harris角点提取的点云粗配准算法[J]. 西安交通大学学报, 2022, 56(3): 33-44. WANG C, TIAN X, GUO R, et al. Coarse point cloud registration based on adaptive Harris corner extraction[J]. Journal of Xi'an Jiaotong University, 2022, 56(3): 33-44. (in Chinese)
[21]	叶雅欣, 王佳盛, 吴烽云, 等. 消毒机器人目标识别定位与包围盒优化[J]. 激光与光电子学进展, 2022, 59(4): 346-354. YE Y X, WANG J S, WU F Y, et al. Target recognition and localization, bounding box optimization of disinfection robot[J]. Laser & Optoelectronics Progress, 2022, 59(4): 346-354. (in Chinese)
[22]	赵洲, 黄攀峰, 陈路. 1种融合卡尔曼滤波的改进时空上下文跟踪算法[J]. 航空学报, 2017, 38(2): 274-284. ZHAO Z, HUANG P F, CHEN L. An improved spatiotemporal context tracking algorithm fused with Kalman filter[J]. Journal of Aeronautics, 2017, 38(2): 274-284. (in Chinese)
[23]	DOSOVITSKIY A, ROS G, CODEVILLA F, et al. CARLA: An open urban driving simulator[C]. The First Annual Conference on Robot. Machine Learning, Mountain View, California: PMLR, 2017.