为研究自动驾驶车辆换道至专用道的过程,针对专用道入口进行交安设计,如图1所示。借鉴高速公路作业区起点的设置方法,在路侧部署专用道入口距离标志、禁止超车标志等,标志设计符合《道路交通标志和标线第2部分:道路交通标志》(GB 5768.2—2009)。考虑车路协同场景,部署路侧单元(Road side unit,RSU),RSU含有通信模块、边缘计算模块和车辆状态监测系统,能够实现车路通信、提供主动管控决策、监测一定范围内车辆的位置和速度。RSU的部署需综合考虑布设成本和网络的连通性等因素,在混合交通流场景下,以通信范围为400 m为标准,设置RSU的平均布设间隔为1200 m[22]。为了保证能够全面监测专用道入口处的车辆状态,在RSU间增设一套雷视设备。
设计长度为的检测区和长度为的执行区。检测区处监测车辆位置和速度信息,以便支持换道信号控制方案的生成。执行区处自动驾驶车辆受信号控制并接受车路(Vehicle to infrastructure,V2I)通信下发的是否可换道指令。在执行区末端设立龙门架,横梁上安装信号灯并显示倒计时。当信号为绿灯时,CAV可换道至内侧专用道;当信号为红灯时,禁止CAV换道至专用道。执行区的长度为缓冲距离与行动距离之和,缓冲距离提供给车辆调整其行车状态的空间,行动距离满足车辆换道所需的距离,如下所示[23]:
为尽可能多地展示交通流信息,对检测区采取离散交通状态编码(Discrete traffic state encoding,DTSE)[28],将路网网格化处理,离散的网络间隔分别为车辆的长度和宽度。生成的位置矩阵作为状态空间的信息输入,其状态空间的大小为。矩阵中每个元素为刻画交通状态的指标,为强化CAV在路网中的重要性,每个元胞取值为。以车辆中心点落在元胞表示对该元胞的占用:0代表元胞无车辆占用;0.5代表有人工驾驶车辆占用;1代表有CAV占用,如图2所示。
策略求解采用DQN算法,结合深度学习特征感知能力和强化学习的决策能力,可解决较复杂的问题。Agent结构采用卷积神经网络(Convolutional neural networks,CNN),包括2层卷积层、1层激励层、1层池化层和1层全连接层,隐藏层激活函数为线性修正单元(Rectified linear units,ReLU)函数,池化层采用最大池化。Agent算法结构如图3所示,输出为Q表中的值。
QinYan-yan, WangHao, WangWei. LWR model for mixed traffic flow in connected and autonomous vehicular environments[J]. China Journal of Highway and Transport, 2018, 31(11): 147-156.
[3]
BeckerF, AxhausenK W. Literature review on surveys investigating the acceptance of automated vehicles[J]. Transportation, 2017, 44(6): 1293-1306.
[4]
ElliottD, KeenW, MiaoL. Recent advances in connected and automated vehicles[J]. Journal of Traffic and Transportation Engineering (English Edition), 2019, 6(2): 109-131.
DuYu-chuan, LiuCheng-long, WuDi-fei, et al. Framework of the new generation of smart highway[J]. China Journal of Highway and Transport, 2022, 35(4): 203-214.
[7]
ShladoverS E. Connected and automated vehicle systems: introduction and overview[J]. Journal of Intelligent Transportation Systems, 2017, 22(3): 190-199.
RanBin, TanHua-chun, ZhangJian, et al. Development status and trend of connected automated vehicle highway system[J]. Journal of Automotive Safety and Energy, 2018, 9(2): 119-130.
[10]
MaK, WangH. Influence of exclusive lanes for connected and autonomous vehicles on freeway traffic flow[J]. IEEE Access, 2019, 7: 50168-50178.
[11]
HuaX D, YuW J, WangW. Influence of lane policies on freeway traffic mixed with manual and connected and autonomous vehicles[J]. Journal of Advanced Transportation, 2020, 2020: No.3968625.
[12]
GhiasiA, HussainO, QianZ. A mixed traffic capacity analysis and lane management model for connected automated vehicles: a markov chain method[J]. Transportation Research Part B, 2017, 106: 266-292.
[13]
ChenD J, AhnS, ChitturiM. Towards vehicle automation: roadway capacity formulation for traffic mixed with regular and automated vehicles[J]. Transportation Research Part B, 2017, 100: 196-221.
[14]
YeL, YamamotoT. Impact of dedicated lanes for connected and autonomous vehicle on traffic flow throughput[J]. Physica A, 2018, 512: 588-597.
[15]
YeL, YamamotoT. Modeling connected and autonomous vehicles in heterogeneous traffic flow[J]. Physica A, 2018, 490: 269-277.
[16]
MasherD P, RossD W, WongP J, et al. Guidelines for design and operation of ramp control systems[R]. Menlo Park: CA United States: Stanford Research Institute,1975.
[17]
SmaragdisE, PapageogiouM. Series of new local ramp metering strategies[J]. Transportation Research Record, 2003, 1856: 74-86.
[18]
BellemansT, SchutterB D, MoorB D. Model predictive control for ramp metering of motorway traffic: a case study[J]. Control Engineering Practice, 2006, 14(7): 757-767.
ChenDe-wang, WangFei-yue, ChenLong. Freeway ramp control algorithm based on neurofuzzy networks[J]. Journal of Traffic and Transportation Engineering, 2003(2): 100-105.
ZengYun-cheng, ShaoMin-hua, SunLi-jun, et al. Traffic prediction and congestion control based on directed graph convolution neural network[J]. China Journal of Highway and Transport, 2021, 34(12): 239-248.
[23]
ZhangJ, JiangX, LiuZ, et al. A study on autonomous intersection management: planning-based strategy improved by convolutional neural network[J]. KSCE Journal of Civil Engineering, 2021, 25(10): 3995-4004.
[24]
DavarynejadM, HegyiA, VranckenJ, et al. Motorway ramp-metering control with queuing consideration using Q-learning[C]//2011 14th International IEEE Conference on Intelligent Transportation Systems (ITSC), Washington, DC, USA, 2011: 1652-1658.
ZhengSi. Research on deep reinforcement learning-based active traffic flow control strategies at distant downstream bottlenecks of expressway[D]. Nanjing: School of Transportation, Southeast University, 2021.
[27]
BellettiF, HazizaD, GomesG. Expert level control of ramp metering based on multi-task deep reinforcement learning[J]. Transactions on Intelligent Transportation Systems, 2017, 19(4): 1198-1207.
[28]
葛家丽. 基于车辆簇的高速公路路侧单元部署研究[D]. 济南: 山东科技大学交通学院, 2020.
[29]
GeJia-li. Research on freeway road side unit deployment based on vehicle clusters[D]. Jinan: School of Transportation, Shandong University of Science and Technology, 2020.
ChenYu. The safety analysis and organization & management method study of freeway work zone[D]. Harbin: School of Transportation Science & Engineering, Harbin Institute of Technology, 2006.
[32]
石茂清. 道路交通安全设施设计研究[D]. 成都: 西南交通大学交通运输与物流学院, 2005.
[33]
ShiMao-qing. Study on design of traffic safety facilities[D]. Chengdu: School of Transportation and Logistics, Southwest Jiaotong University, 2005.
SunLing, ZhangJing, ZhouYing, et al. Design of entrance area of automatic driving special lane in vehicle-infrastructure collaborative environment[J]. Journal of Highway and Transportation Research and Development, 2020, 37(Sup.1): 122-129.
[36]
MilanésV, ShladoverS E. Modeling cooperative and autonomous adaptive cruise control dynamic responses using experimental data[J]. Transportation Research Part C: Emerging Technologies, 2014, 48: 285-300.
[37]
XueY, WangX, CenB L, et al. Study on fuel consumption in the Kerner-Klenov-Wolf three-phase cellular automaton traffic flow model[J]. Nonlinear Dynamics, 2020, 102(1): 1-10.
[38]
WangC H, HwangM C. Value-based deep reinforcement learning for adaptive isolated intersection signal control[J]. IET Intelligent Transport Systems, 2018, 12(9): 1005-1010.
[39]
SchaulT, QuanJ, AntonoglouI, et al. Prioritized experience replay[C]//Proceeding of the 4th International Conference on Learning Representations, San Juan, Puerto Rico, 2016: 322-355.
[40]
武毅. 基于三相交通流理论的元胞自动机模型研究[D]. 长春: 吉林大学交通学院, 2018.
[41]
WuYi. Research on cellular automated traffic flow model for three-phase theory[D]. Changchun: College of Transportation, Jilin University, 2018.
ShuLing-zhou, WuJia, WangChen. Urban traffic signal control based on deep reinforcement learning[J]. Journal of Computer Applications, 2019, 39(5): 1495-1499.