基于神经网络的车辆强制换道预测模型

崔洁茗; 余贵珍; 周彬; 李存金; 马继伟; 徐国艳

doi:10.13700/j.bh.1001-5965.2020.0662

基于神经网络的车辆强制换道预测模型

doi: 10.13700/j.bh.1001-5965.2020.0662

1.
北京航空航天大学交通科学与工程学院特种车辆无人运输技术工业和信息化部重点实验室, 北京 100083
2.
内蒙古霍林河露天煤业股份有限公司, 通辽 028001

基金项目:

国家自然科学基金 51775016

北京市自然科学基金 L191002

详细信息

通讯作者:
周彬, E-mail: binzhou@buaa.edu.cn

中图分类号: TP181;U491.1⁺4
计量
- 文章访问数: 526
- HTML全文浏览量: 188
- PDF下载量: 59
- 被引次数: 0
出版历程
- 收稿日期: 2020-11-26
- 录用日期: 2021-03-05
- 网络出版日期: 2022-05-20

Mandatory lane change decision-making model based on neural network

1.
Key Laboratory of Autonomous Transportation Technology for Special Vehicles, Ministry of Industry and Information Technology, School of Transportation Science and Engineering, Beihang University, Beijing 100083, China
2.
Inner Mongolia Huolinhe Surface Coal Industry Co., Ltd., Tongliao 028001, China

Funds:

National Natural Science Foundation of China 51775016

Beijing Municipal Natural Science Foundation L191002

More Information

Corresponding author: ZHOU Bin, E-mail: binzhou@buaa.edu.cn

摘要

摘要:
针对高速公路上车辆行驶速度快，换道行为危险程度高的问题，聚焦于不可避免、发生频繁、一旦发生事故后果严重的强制换道行为，改进基于门控循环单元(GRU)的换道模型，对强制换道行为进行分析与预测。为保证模型的有效性，选取下一代仿真技术(NGSIM)数据作为模型的训练集与检测集，使用侧向加速度将车辆侧向摆动数据有效删除并得到强制换道的最迟换道点，进而实现车辆位置与换道决策的预测。实验结果证明，所提模型能够以96.01%的准确率判定车辆在最迟换道点的强制换道行为，相较于LSTM模型准确率提升了3.67%，同时相较于朴素贝叶斯网络准确率提高了7.31%。
- 强制换道行为 /
- 神经网络 /
- 换道决策 /
- 侧向加速度 /
- 门控循环单元(GRU)
Abstract:
Aiming at the problem of fast-speed and high risk of lane changing behavior on expressway, we focus on the ineviteable, freguent and serve mandatory lane-changing behaviors to improve the lane-changing model based on gated recurrent unit (GRU), and predict the decision-making behaviors of mandatony lane-changing. To verify the effectiveness of the model, adopt the next generation simulation (NGSIM) data as the training set and test set of the model. From this data, the lateral acceleration threshold is obtained to screen out the phenomenon of lateral swing of vehicles. The experimental results indicate that the optimized model could determine the location of mandatory lane change with an accuracy of 96.01%. The accuracy of the model is improved by 3.67% compared with the LSTM model, and is improved by 7.31% compared with the naive Bayes network.
- mandatory lane change behavior /
- neural networks /
- lane change decision making /
- lateral acceleration /
- gated recurrent unit (GRU)

HTML全文

图 1 强制换道示意图

Figure 1. Schematic diagram of mandatory lane change

下载: 全尺寸图片幻灯片

图 2 换道决策流程

Figure 2. Flowchart of lane change decision making

下载: 全尺寸图片幻灯片

图 3 本文决策模型结构

Figure 3. Structure of proposed decision-making model

下载: 全尺寸图片幻灯片

图 4 本文模型算法流程

Figure 4. Flowchart of proposed model's algorithm

下载: 全尺寸图片幻灯片

图 5 本文模型网络结构

Figure 5. Network structure of proposed model

下载: 全尺寸图片幻灯片

图 6 I-80数据库检测范围示意图

Figure 6. Selected detection area section of I-80 database

下载: 全尺寸图片幻灯片

图 7 I-80数据库整体检测范围示意图

Figure 7. Whole detection area section of I-80 database

下载: 全尺寸图片幻灯片

图 8 数据去噪

Figure 8. Data denoising

下载: 全尺寸图片幻灯片

图 9 车辆5的可能换道点提取

Figure 9. Possible lane change points extraction of vehicle 5

下载: 全尺寸图片幻灯片

图 10 全部车辆可能换道点与真实换道点比较

Figure 10. Comparison between possible and real lane change points of all vehicles

下载: 全尺寸图片幻灯片

图 11 竖直方向车辆合流点聚类

Figure 11. Vertical-direction vehicle confluence points clustering

下载: 全尺寸图片幻灯片

图 12 当前车辆与相邻车道前后车辆间的关系

Figure 12. Relationship among present vehicle and front and back vehicle on adjacent lane

下载: 全尺寸图片幻灯片

图 13 隐藏层层数与神经元个数对模型的影响

Figure 13. Effect of the number of hidden layers and neurons on model

下载: 全尺寸图片幻灯片

表 1 NGSIM数据(部分)

Table 1. NGSIM data (partial)

名称	描述	单位	转换后单位
车辆ID	被检测车辆的序号
时间	统一的车辆被检测的时间	0.1 s	0.1 s
X位置坐标	被检测车辆与道路左侧的距离	ft	m
Y位置坐标	被检测车辆与检测路段起点的距离	ft	m
车辆类别	车辆按大小/型号进行的分类
车辆速度	车辆瞬时速度	ft/s	m/s
车头间距	2辆连续行驶的车辆车头间的距离	ft	m
注：1 ft(英尺)=0.304 8 m。

下载: 导出CSV

表 2 强制换道车辆数据(部分)

Table 2. Mandatory lane change vehicle data (partial)

车辆ID	时间戳	X位置坐标/m	Y位置坐标/m	车辆速度/(m·s^-1)	车辆加速度/(m·s^-2)	车道序号
5	1113433148400	68.874	65.907	21.55	0	6
5	1113433148500	68.875	68.076	21.55	0	6
5	1113433150300	67.414	108.23	24.83	-0.7	6
⋮	⋮	⋮	⋮	⋮	⋮	⋮
2 911	1113433959100	74.76	679.233	15.91	-0.16	6
2 911	1113433959200	74.665	680.823	15.88	0.02	6

下载: 导出CSV

表 3 车辆5的强制换道数据

Table 3. Mandatory lane change data of vehicle 5

车辆ID	X位置坐标/m	Y位置坐标/m	车辆速度/(m·s^-1)	车辆加速度/(m·s^-2)	车道序号
1	69.557	303.541	7.08	-1.19	6
2	69.547	305.518	20.16	-0.1	6
⋮	⋮	⋮	⋮	⋮	⋮
211	74.423	697.387	19.55	-1.09	7
212	74.314	699.343	19.52	-0.43	7

下载: 导出CSV

表 4 不同学习率的MAE对比

Table 4. Comparison of MAE with different learning rates

学习率	MAE
0.1	2.673 574 441 2
0.01	0.086 840 446 7
0.001	0.061 312 660 5
0.000 1	1.105 762 974 3

下载: 导出CSV

表 5 强制换道决策模型预测结果

Table 5. Prediction result of mandatory lane change decision-making model

车辆ID	X位置坐标/m	Y位置坐标/m	车辆速度/(m·s^-1)	侧向加速度/(m·s^-2)	真实换道情况	预测换道情况(bool)
5	76.331	661.762	22.09	1.67	1	True
⋮	⋮	⋮	⋮	⋮	⋮	⋮
186	76.329	664.643	18.86	1.63	1	True
⋮	⋮	⋮	⋮	⋮	⋮	⋮
2 911	74.886	664.511	16.09	1.75	1	True
2 911	74.76	679.233	15.91	-0.16	0	False
2 911	74.665	680.823	15.88	0.02	0	False

下载: 导出CSV

表 6 改进前后模准确率对比

Table 6. Comparison of model accuracy before and after improvement

模型	准确率/%	召回率/%
强制换道决策模型	96.01	98.58
GRU模型	91.73	100
LSTM模型^[20]	92.34	99.32
朴素贝叶斯模型^[14]	88.70

下载: 导出CSV

参考文献(21)

[1]	WOLFRAM S. Theory and application of cellular automata[M]. Singapore: World Scientific, 1986.
[2]	NAGEL K, SCHRECKENBERG M. A cellular automaton model for freeway traffic[J]. Journal De Physique I, 1992, 2(12): 2221-2229. doi: 10.1051/jp2:1992262
[3]	CHOWDHURY D, WOLF D E, SCHRECKENBERG M. Particle hopping models for two-lane traffic with two kinds of vehicles: Effects of lane-changing rules[J]. Physics A: Statistical Mechanics and its Applications, 1997, 235(3): 417-439.
[4]	GIPPS P G. A model for the structure of lane-changing decisions[J]. Transportation Research Part B: Methodological, 1986, 20(5): 403-414. doi: 10.1016/0191-2615(86)90012-3
[5]	LEE G. Modeling gap acceptance at freeway merges[D]. Cambridge: MIT, 2006.
[6]	HIDAS P. Modeling lane changing and merging in microscopic traffic simulation[J]. Transportation Research Part C: Emerging Technologies, 2002, 10(5): 351-371.
[7]	BOXILL S A, YU L. An evaluation of traffic simulation models for supporting ITS[D]. Houston: Texas Szouthern University, 2000.
[8]	TALEBPOUR A, MAHMASSANI H S, HAMDAR S H. Modeling lane-changing behavior in a connected environment: A game theory approach[J]. Transportation Research Procedia, 2015, 7: 420-440. doi: 10.1016/j.trpro.2015.06.022
[9]	MENG F L, SU J Y, LIU C J, et al. Dynamic decision making in lane change: Game theory with receding horizon[C]//2016 UKACC 11th International Conference on Control. Piscataway: IEEE Press, 2016: 1-6.
[10]	NILSSON J, SJÖBERG J. Strategic decision making for automated driving on two-lane, one way roads using model predictive control[C]//2013 Intelligent Vehicles Symposium (IV). Piscataway: IEEE Press, 2013: 1253-1258.
[11]	WANG Y, ZHANG D, LIU Y, et al. Enhancing transportation systems via deep learning: A survey[J]. Transportation Research Part C: Emerging Technologies, 2019, 99: 144-163. doi: 10.1016/j.trc.2018.12.004
[12]	DING C, WANG W, WANG X, et al. A neural network model for driver's lane-changing trajectory prediction in urban traffic flow[J]. Mathematical Problems in Engineering, 2013, 2013: 967358.1-967358.8.
[13]	LIU Y, WANG X, LI L, et al. A novel lane change decision-making model of autonomous vehicle based on support vector machine[J]. IEEE Access, 2019, 7: 26543-26550. doi: 10.1109/ACCESS.2019.2900416
[14]	邱小平, 刘亚龙, 马丽娜, 等. 基于贝叶斯网络的车辆换道模型[J]. 交通运输系统工程与信息, 2015, 15(5): 67-73. https://www.cnki.com.cn/Article/CJFDTOTAL-YSXT201505010.htm QIU X P, LIU Y L, MA L N, et al. Vehicle lane change model based on Bayesian network[J] Transportation Systems Engineering and Information, 2015, 15(5): 67-73(in Chinese). https://www.cnki.com.cn/Article/CJFDTOTAL-YSXT201505010.htm
[15]	HUANG X, SUN J, SUN J. A car-following model considering asymmetric driving behavior based on long short-term memory neural networks[J]. Transportation Research Part C: Emerging Technologies, 2018, 95: 346-362. doi: 10.1016/j.trc.2018.07.022
[16]	HOU Y, EDARA P, SUN C. Modeling mandatory lane changing using Bayes classifier and decision trees[J]. Intelligent Transportation Systems, 2014, 15(2): 647-655.
[17]	ZHENG J, SUZUKI K, FUJITA M. Predicting driver's lane-changing decisions using a neural network model[J]. Simulation Modelling Practice and Theory, 2014, 42: 73-83. doi: 10.1016/j.simpat.2013.12.007
[18]	CHO K, VAN MERRIENBOER B, GULCEHRE C, et al. Learning phrase representations using RNN encoder-decoder for statistical machine translation[EB/OL]. (2014-09-03)[2020-11-01]. https://arxiv.org/ahs/1406.1078.
[19]	FU R, ZHANG Z, LI L. Using LSTM and GRU neural network methods for traffic flow prediction[C]//2016 31st Youth Academic Annual Conference of Chinese Association of Automation (YAC). Piscataway: IEEE Press, 2016: 324-328.
[20]	XIE D F, FANG Z Z, JIA B, et al. A data-driven lane-changing model based on deep learning[J]. Transportation Research Part C: Emerging Technologies, 2019, 106: 41-60. doi: 10.1016/j.trc.2019.07.002
[21]	吴江玲, 张生瑞, AMIT K S, 等. 高速公路强制换道持续时间半参数生存分析[J]. 浙江大学学报(工学版), 2017, 51(11): 2215-2221. doi: 10.3785/j.issn.1008-973X.2017.11.016 WU J L, ZHANG S R, AMIT K S, et al. Semi-parametric survival analysis of mandatory lane changing duration of freeways[J] Journal of Zhejiang University(Engineering Science), 2017, 51(11): 2215-2221(in Chinese). doi: 10.3785/j.issn.1008-973X.2017.11.016