Methods for predicting vehicle trajectories in motorway weaving zones based on driving risk fields

doi:10.3969/j.issn.1674-8484.2024.06.017

Abstract

Abstract:

A vehicle trajectory prediction method fusing the traveling risk field and vehicle lane-changing intention was proposed to improve the accuracy of vehicle trajectory prediction in the interweaving area. Firstly, the driving demand changes of drivers in the interweaving zone were analyzed, and the driving risk field model was used to uniformly represent the interaction risk when vehicles were driving; secondly, the Hidden Markov Model was used to identify the vehicle lane-changing intention; in addition, the input features were extended and fused in multiple dimensions by the Deep Belief Networks Online Learning Machine (DBN_OSELM) model, to improve the accuracy of the trajectory prediction in the interweaving zone. Finally, the proposed method was evaluated based on the CitySim dataset. The results show that the model can predict vehicle trajectories in the interweaving zone of highways with high accuracy, and the root mean square error (RMSE) of vehicle trajectory prediction for the three types (merging in the confluence area, maintaining the interweaving area, and driving out the diverging area) of driving needs of drivers in the interweaving zone are 0.683 5, 0.257 4, and 0.631 5, respectively, and the average displacement error (ADE) is 0.46, 0.21, and 0.48 m, respectively. The research results are helpful to improve the accuracy of vehicle trajectory prediction in complex scenarios and improve traffic safety in the intertwined area.

Key words: intelligent transportation, driving demand, driving risk field, lane-change intention, deep belief networks online learning machine (DBN_OSELM) model, trajectory prediction

CLC Number:

U492.8+4

QIN Yaqin, DONG Shuai, XIE Jiming, CHEN Liang, LIU Yonghua, GUO Miao. Methods for predicting vehicle trajectories in motorway weaving zones based on driving risk fields[J]. Journal of Automotive Safety and Energy, 2024, 15(6): 952-961.

Figures/Tables 15

References 21

[1]	Vishnu C, Abhinav V, Roy D, et al. Improving multi-agent trajectory prediction using traffic states on interactive driving scenarios[J]. IEEE Robot Auto Letts, 2023, 8(5): 2708-2715.
[2]	殷子健, 徐良杰, 刘伟, 等. 基于迁移学习的高速公路交织区车辆轨迹预测[J]. 深圳大学学报(理工版), 2024, 41(1): 92-100.
	YIN Zijian, XU Lujie, LIU Wei, et al. Vehicle trajectory prediction in weaving area of expressway based on transfer learning[J]. J Shenzhen Univ (Sci Engi), 2024, 41(1): 92-100. (in Chinese)
[3]	Khatib O. Real-time obstacle avoidance for manipulators and mobile robots[J]. Int’l J Robot Res, 1986, 5(1): 90-98.
[4]	Wolf M T, Burdick J W. Artificial potential functions for highway driving with collision avoidance[C]// 2008 IEEE Int’l Conf Robot Autom. IEEE, 2008: 3731-3736.
[5]	王建强, 吴剑, 李洋. 基于人-车-路协同的行车风险场概念、原理及建模[J]. 中国公路学报, 2016, 29(1): 105-114.
	WANG Jianqiang, WU Jian, LI Yang. Concept, principle and modeling of driving risk field based on driver-vehicle-road interaction[J]. Chin J Highw Transport, 2016, 29(1): 105-114. (in Chinese)
[6]	李林恒, 甘婧, 曲栩, 等. 智能网联环境下基于安全势场理论的车辆跟驰模型[J]. 中国公路学报, 2019, 32(12): 76-87. doi: 10.19721/j.cnki.1001-7372.2019.12.008
	LI Linhen, GAN Jing, QU Xu, et al. Car-following model based on safety potential field theory under connected and automated vehicle environment[J]. Chin J Highw Transport, 2019, 32(12): 76-87. (in Chinese) doi: 10.19721/j.cnki.1001-7372.2019.12.008
[7]	李林恒, 甘婧, 曲栩, 等. 智能网联环境下基于安全势场理论的车辆换道模型[J]. 中国公路学报, 2021, 34(6): 184-195. doi: 10.19721/j.cnki.1001-7372.2021.06.018
	LI Linhen, GAN Jing, QU Xu, et al. Lane-changing model based on safety potential field theory under the connected and automated vehicles environment[J]. Chin J Highw Transport, 2021, 34(6): 184-195. (in Chinese) doi: 10.19721/j.cnki.1001-7372.2021.06.018
[8]	HUANG Yanjun, DU Jiatong, YANG Ziru, et al. A survey on trajectory-prediction methods for autonomous driving[J]. IEEE Trans Intel Vehi, 2022, 7(3): 652-674.
[9]	王明, 唐小林, 杨凯, 等. 考虑预测风险的自动驾驶车辆运动规划方法[J]. 汽车工程, 2023, 45(8): 1362-1372, 1407.
	WANG Ming, TANG Xiaolin, YAN Kai, et al. A motion planning method for autonomous vehicles considering prediction risk[J]. Autom Engineering, 2023, 45(8): 1362-1372, 1407. (in Chinese)
[10]	高镇海, 鲍明喜, 高菲, 等. 基于LSTM概率多模态预期轨迹预测方法[J]. 汽车工程, 2023, 45(7): 1145-1152.
	GAO Zhenhai, BAO Mingxi, GAO Fei, et al. The method of probabilistic multi-modal expected trajectory prediction based on LSTM[J] Autom Engineering, 2023, 45(7):1145-1152. (in Chinese)
[11]	MO Xiaoyu, XING Yang, LYU Chen. Graph and recurrent neural network based vehicle trajectory prediction for highway driving [C]// 2021 IEEE Int’l Intel Transport Syst Conf (ITSC). IEEE, 2021: 1934-1939.
[12]	WANG Xiao, TANG Ke, DAI Xingyuan, et al. Safety-balanced driving-style aware trajectory planning in intersection scenarios with uncertain environment[J]. IEEE Trans Intel Vehi, 2023, 8(4): 2888-2898.
[13]	金立生, 魏青嵩, 谢宪毅, 等. 基于DMPC的无信控交叉口智能网联车辆多车协同轨迹规划[J]. 汽车安全与节能学报, 2024, 15(2): 235-241.
	JIN Lisheng, WEI Qingsong, XIE Xianyi, et al, Multi-vehicle cooperative path planning at untrusted intersections based on DMPC[J]. J Autom Safe Energ, 2024, 15(2): 235-241. (in Chinese)
[14]	宋秀兰, 董兆航, 单杭冠, 等. 基于时空融合的多头注意力车辆轨迹预测[J]. 浙江大学学报(工学版), 2023, 57(8): 1636-1643.
	SONG Xiulan, DONG Zhaohan, SHANG Hanguan. Vehicle trajectory prediction based on temporal-spatialmulti-head attention mechanism[J]. J Zhejiang Univ (Engi Sci), 2023, 57(8): 1636-1643. (in Chinese)
[15]	冯玉荣, 杨少伟, 杨宏志, 等. 高速公路交织区运行分析方法对比研究[J]. 公路交通科技, 2018, 35(8): 134-143. doi: 10.3969/j.issn.1002-0268.2018.08.018
	FENG Yurong, YANG Shaowei, YANG Hongzhi, et al. Comparative study on operation analysis methods of expressway weaving area[J]. J Highw Transport Res Develop, 2018, 35(8): 134-143. (in Chinese)
[16]	Mantouka E G, Barmpounakis E N, Vlahogianni E I. Identifying driving safety profiles from smartphone data using unsupervised learning[J]. Safe Sci, 2019, 119: 84-90.
[17]	Mehrabian A R, Lucas C. A novel numerical optimization algorithm inspired from weed colonization[J]. Ecolog Info, 2006, 1(4): 355-366.
[18]	SONG Xianmin, SUN Yaqian, TAN Pengfei. A dynamic bayesian network model for real-time risk propagation of secondary rear-end collision accident using driving risk field[J]. IEEE Access, 2022, 10: 72429-72443.
[19]	胡春燕, 曲大义, 赵梓旭, 等. 考虑前车驾驶风格的改进自适应巡航控制跟驰模型及仿真[J]. 济南大学学报(自然科学版), 2023, 37(3): 331-338.
	HU Chunyana, QU Dayi, ZHAO Zixu, et al. lmproved Adaptive Cruise Control Car-following Model and Simulation Considering Driving Styles of Leading Car[J]. J Jinan Univ (Nat Sci Edit), 2023, 37(3): 331-338. (in Chinese)
[20]	YUAN Renteng, Abdel-Aty M, GU Xin, et al. A unified approach to lane change intention recognition and driving status prediction through TCN-LSTM and multi-task learning models[J]. Phys A: Statist Mech its Appl, 2023, 632(1): 129-332.
[21]	谢济铭, 秦雅琴, 彭博, 等. 多车道交织区车辆跟驰行为风险判别与冲突预测[J]. 交通运输系统工程与信息, 2021, 21(3): 131-139.
	XIE Jiming, QIN Yaqin, PENG Bo, et al. Risk discrimination and conflict prediction of vehicle-following behavior in multi-lane weaving sections[J]. J Transport Syst Engi Info Tech, 2021, 21(3): 131-139. (in Chinese)

聚类算法	MAE	MSE
GMM	0.849 5	1.299 3
IWO-K-means	0.162 7	0.263 0
K-Means	0.824 6	1.202 9

聚类算法	MAE	MSE
GMM	0.849 5	1.299 3
IWO-K-means	0.162 7	0.263 0
K-Means	0.824 6	1.202 9

驾驶需求	速度平均值 (m·s^-1)	速度标准差 (m·s^-1)	加速度平均值 (m·s^-2)	加速度标准差 (m·s^-2)	减速度平均值 (m·s^-2)	减速度标准差 (m·s^-2)
合流段汇入	16.86	5.11	3.69	2.66	-3.71	2.64
交织段保持	16.57	5.21	3.63	2.57	-3.4	2.55
分流段驶出	17.01	4.83	3.95	2.94	-3.97	2.92

驾驶需求	速度平均值 (m·s^-1)	速度标准差 (m·s^-1)	加速度平均值 (m·s^-2)	加速度标准差 (m·s^-2)	减速度平均值 (m·s^-2)	减速度标准差 (m·s^-2)
合流段汇入	16.86	5.11	3.69	2.66	-3.71	2.64
交织段保持	16.57	5.21	3.63	2.57	-3.4	2.55
分流段驶出	17.01	4.83	3.95	2.94	-3.97	2.92

模型	风险场	驾驶意图	历史轨迹	RMSE	ADE / m
A	-	-	√	0.982 0	0.77
B	√	-	√	0.794 7	0.57
C	-	√	√	0.681 3	0.54
D	√	√	√	0.552 6	0.43