|Table of Contents|

Recognition of disturbed trajectories of right-turning vehicles at signalized intersections based on trajectory multi-feature agglomerative hierarchical clustering and Gaussian mixture model(PDF)

长安大学学报（自然科学版）[ISSN:1006-6977/CN:61-1281/TN]

Issue:
2025年2期
Page:
136-153
Research Field:
交通工程
Publishing date:

Info

Title:
Recognition of disturbed trajectories of right-turning vehicles at signalized intersections based on trajectory multi-feature agglomerative hierarchical clustering and Gaussian mixture model
Author(s):
LIANG Guo-hua1;  ZHANG Yu-jie1;  CHEN Yi-xin1;  WENG Kang-yi2; SUN Run-feng3;  MENG Xiao-yang1
(1. School of Transportation Engineering, Chang'an University, Xi'an 710064, Shaanxi, China; 2. School of Information Engineering, Chang'an University, Xi'an 710018, Shaanxi, China; 3. School of Civil Engineering, Beijing Jiaotong University, Beijing 100044, China)
Keywords:
traffic engineering;  vehicle trajectory;  trajectory multi-feature agglomerative hierarchical clustering;  Gaussian mixture model(GMM);  anomaly recognition
PACS:
U491
DOI:
10.19721/j.cnki.1671-8879.2025.02.012
Abstract:
To enhance the capability of right-turning vehicle drivers in responding to interference factors such as traffic signal phasing, pedestrians, and non-motorized vehicles, real-world trajectory data of right-turning vehicles at signalized intersections were utilized. Trajectory multi-feature agglomerative hierarchical clustering based on principal direction and Hausdorff distance was implemented to classify distinct trajectory operation patterns. Statistical analysis of disturbance-induced variations in trajectory characteristics was conducted to determine disturbance thresholds for speed and steering angle features. A Gaussian mixture model(GMM)was employed by integrating the disturbance-thresholds with clustering results to detect disturbed anomaly trajectories. The research results show that the enhanced agglomerative hierarchical clustering algorithm based on the trajectory principal direction and Hausdorff distance can effectively characterize distinct trajectory motion patterns. The trajectory speed and steering angle can be used as criteria to determine whether the trajectory is disturbed, where the speed disturbance threshold is defined as the mean minus two times standard deviations, and the steering angle disturbance threshold is set at the mean plus or minus two times standard deviations. The trajectory multi-feature agglomerative hierarchical clustering and Gaussian mixture model is proven can effectively screen the disturbed right-turning trajectories, with a recognition accuracy of 99.95%. Thus the proposed method can accurately identify anomaly trajectories. This framework serves as a theoretical foundation for assessing driving risks in right-turning maneuvers and designing driving assistance systems, while simultaneously providing early warning mechanisms to ensure the driving safety of vehicles at intersections.5 tabs, 23 figs, 33 refs.

References:

[1] 唐克双,杨博文,许 凯,等.基于车辆轨迹数据的交叉口危险驾驶行为预测[J].同济大学学报(自然科学版),2017,45(10):1454-1461.
TANG Ke-shuang,YANG Bo-wen,XU Kai,et al.Intersection hazardous driving behaviorprediction based on vehicle trajectorydata[J].Journal of Tongji University(Natural Science Edition),2017,45(10):1454-1461.
[2]白乔文,曲昭伟,陈永恒,等.非严格优先权下无左转专用相位直行车辆轨迹模型建立[J].吉林大学学报(工学版),2019,49(3):673-679.
BAI Qiao-wen,QU Zhao-wei,CHEN Yong-heng,et al.Establishment of trajectorymodel for straight-going vehicles withoutdedicated left-turn phase under non-strict priority[J].Journal of Jilin University(Engineering and Technology Edition),2019,49(3):673-679.
[3]高雪婷.基于深度学习的行人和骑行者检测及防碰撞预警系统研究[D].镇江:江苏大学,2018.
GAO Xue-ting.Research on pedestrian andcyclist detection and collision avoidancewarning system based on deep learning[D].Zhenjiang:Jiangsu University,2018.
[4]甘元震.高速公路车辆异常行驶状态预警研究[D].石家庄:石家庄铁道大学,2019.
GAN Yuan-zhen.Research on early warningof abnormal driving states of vehicles onexpressways[D].Shijiazhuang:Shijiazhuang TiedaoUniversity,2019.
[5]倪 捷,刘志强,涂孝军,等.面向驾驶辅助系统的换道安全性预测模型研究[J].交通运输系统工程与信息,2016,16(4):95-100.
NI Jie,LIU Zhi-qiang,TU Xiao-jun,et al.Research on lane-changing safetyprediction model for driving assistancesystems[J].Journal of TransportationSystems Engineering and InformationTechnology,2016,16(4):95-100.
[6]牟乃夏,徐玉静,张恒才,等.移动轨迹聚类方法研究综述[J].测绘通报,2018(1):1-7.
MU Nai-xia,XU Yu-jing,ZHANG Heng-cai,et al.Review of research on mobile trajectoryclustering methods[J].Bulletin of Surveyingand Mapping,2018(1):1-7.
[7]韩 晨.面向轨迹大数据的高效聚类算法设计与实现[D].呼和浩特:内蒙古大学,2021.
HAN Chen.Design and implementation ofefficient clustering algorithm fortrajectory big data[D].Hohhot:Inner Mongolia University,2021.
[8]孙石磊,王 超,赵元棣.基于轮廓系数的参数无关空中交通轨迹聚类方法[J].计算机应用,2019,39(11):3293-3297.
SUN Shi-lei,WANG Chao,ZHAO Yuan-di.Aparameter-independent air traffictrajectory clustering method based on thesilhouette coefficient[J].Journal ofComputer Applications,2019,39(11):3293-3297.
[9]KUMAR D,WU H,RAJASEGARAR S,et al.Fast and scalable big data trajectoryclustering for understanding urbanmobility[J].IEEE Transactions onIntelligent Transportation Systems,2018,19(11):3709-3722.
[10]NIU X,CHEN T,WU C Q,et al.Label-based trajectory clustering in complexroad networks[J].IEEE Transactionson Intelligent Transportation Systems,2019,21(10):4098-4110.
[11]WANG C,LI J,HE Y,et al.Segmentedtrajectory clustering-based destinationprediction in IoVs[J].IEEE Access,2020,8:98999-99009.
[12]HAN Y,TSE R,CAMPBELL M.Pedestrianmotion model using non-parametrictrajectory clustering and discretetransition points[J].IEEE Robotics andAutomation Letters,2019,4(3):2614-2621.
[13]AILIN H,ZHONG L,DECHAO Z.Movementpattern extraction based on a non-parameter sub-trajectory clusteringalgorithm[C]//IEEE.2019 IEEE 4thInternational Conference on Big DataAnalytics(ICBDA).New York:IEEE,2019:5-9.
[14]CHOONG M Y,ANGELINE L,CHIN R K Y,et al.Modeling of vehicle trajectory usingK means and fuzzy C-means clustering[C]//IEEE.2018 IEEE InternationalConference on Artificial Intelligence inEngineering and Technology(IICAIET).New York:IEEE,2018:1-6.
[15]SALARPOUR A,KHOTANLOU H.Direction-based similarity measure to trajectoryclustering[J].IET Signal Processing,2019,13(1):70-76.
[16]ALAHI A,GOEL K,RAMANATHAN V,et al.Social LSTM:Human trajectoryprediction in crowded spaces[C]//IEEE.2016 IEEE Conference onComputer Vision and PatternRecognition(CVPR),NewYork:IEEE,2016:961-971.
[17]王金凤,邢长征.基于道路网络的移动对象聚类[J].计算机工程与应用,2016,52(7):74-78.
WANG Jin-feng,XING Chang-zheng.Clustering of mobile objects based on roadnetwork[J].Journal of Computer Engineeringand Applications,2016,52(7):74-78.
[18]袁 冠,夏士雄,张 磊,等.基于结构相似度的轨迹聚类算法[J].通信学报,2011,32(9):103-110.
YUAN Guan,XIA Shi-xiong,ZHANG Lei,et al.Trajectory clustering algorithm based onstructural similarity[J].Journal onCommunications,2011,32(9):103-110.
[19]ATEV S,MILLER G,PAPANIKOLOPOULOS N P.Clustering of vehicle trajectories[J].IEEE Transactions on Intelligent Transportation Systems,2010,11(3):647-657.
[20]ZUCCHINI W,MACDONALD I L,LANGROCK R.Hidden Markov models for time series:An introduction using R[M].Boca Raton:Chapman & Hall/CRCMonographs on Statistics and AppliedProbability,2017.
[21]WANG P,YANG J,ZHANG J.A spatial-contextual indoor trajectory predictionapproach via hidden Markov models[J].Wireless Communications andMobile Computing,2022,2022:1-13.
[22]MORRIS B T,TRIVEDI M M.Learning,modeling,and classification of vehicletrack patterns from live video[J].IEEETransactions on Intelligent Transportation Systems,2008,9(3):425-437.
[23]ABIODUN O I,JANTAN A,OMOLARA A E,et al.State-of-the-art in artificial neuralnetwork applications:A survey[J].Heliyon,2018,4(11):e00938.
[24]JANET J P,DUAN C,YANG T,et al.A quantitative uncertainty metriccontrols error in neural network-driven chemical discovery[J].Chemicalscience,2019,10(34):7913-7922.
[25]GUO L,QIN Z,GE P,et al.Adaptive lane-departure prediction method withsupport vector machine and gated recurrent unit models[J].Journal ofTransportation Engineering,Part A:Systems,2022,148(11):04022103.
[26]GHOSH S,DASGUPTA A,SWETAPADMA A.A study on support vector machinebased linear and non-linear pattern classification[C]//IEEE.2019 International Conference on Intelligent Sustainable Systems(ICISS).New York:IEEE,2019:24-28.
[27]GOLI S A,FAR B H,FAPOJUWO A O.Vehicle trajectory prediction with Gaussianprocess regression in connectedvehicle environment[C]//IEEE.2018IEEE Intelligent Vehicles Symposium(IV).New York:IEEE,2018:550-555.
[28]KIM E,CHOI E.Critical aggressive acceleration values and models for fuel consumption when starting and driving a passenger car running on LPG[J].International Journal of Sustainable Transportation. 2017,11(6):395-405.
[29]ALTCHE F,FORTELLE A D L.An LSTMnetwork for highway trajectoryprediction[C]//IEEE.IEEEInternational Conference on IntelligentTransportation Systems(ITSC).NewYork:IEEE,2017:354-359.
[30]LI F,TANG K,YAO J,et al.Real-timequeue length estimation for signalizedintersections using vehicle trajectorydata[J].Transportation ResearchRecord,2017(2623):49-59.
[31]戴振华,廖祺硕,潘存书,等.城市干路交叉口右转车辆轨迹流线与曲率特性分析[J].交通信息与安全,2022,40(4):157-166.
DAI Zhen-hua,LIAO Qi-shuo,PAN Cun-shu,et al.Analysis of trajectory streamlines andcurvature characteristics of right-turningvehicles at urban arterial roadintersections[J].Journal of TransportInformation and Safety,2022,40(4):157 -166.
[32]倪 颖,辛榴燕,龙柯宇.机非交互的交叉口右转机动车行为建模与仿真[J].同济大学学报(自然科学版),2019,47(7):994-1003.
NI Ying,XIN Liu-yan,LONG Ke-yu.Modelingand simulation of non-cooperative right-turning vehicle behavior at intersections[J].Journal of Tongji University(NaturalScience Edition),2019,47(7):994-1003.
[33]曲昭伟,罗瑞琪,陈永恒,等.信号交叉口右转机动车轨迹特性[J].浙江大学学报(工学版),2018,52(2):341-351.
QU Zhao-wei,LUO Rui-qi,CHEN Yong-heng,et al.Trajectory characteristics of right-turning vehicles at signalized intersections[J].Journal of Zhejiang University(Engineering Science),2018,52(2):341-351.

Memo

Memo:
-
Last Update: 2025-04-01