Analysis of cooperative driving strategies at road network level with macroscopic fundamental diagram

Abstract

本研究对智能网联汽车协同决策问题开展了理论研究,分析了协同决策的内在机理,揭示了不同协同决策算法之间的本质区别,构建了以宏观基本图为核心的评估方法,验证了协同决策的瓶颈要素:二维路权冲突,提出了统一的拉链式协同决策算法框架。该研究为协同决策技术的进一步发展指明了方向,具有重要意义。

Publication
Transportation Research Part C: Emerging Technologies
(中科院1区TOP期刊; JCR Q1区; 影响因子=9.551)

下面为本研究的简要介绍,详细内容请参阅论文原文

A typical isolated intersection and an illustration of the passing order. Vehicles A, B, C, and F will pass through the intersection. Order AFBCmeans that A, F, B, and C will pass through the conflict zone successively. In particular, it should be noted that some vehicles do not have conflicting paths, and thus they can drive into conflict zone simultaneously. Therefore, the order AFBC and FABC are essentially equivalent.


Structure of a well-defined MFD


Overview of the simulation platform


A local intersection in the traffic network, where intersection conflict zone is decomposed to improve space utilization.


The simulated two-lane road network with four intersections.


MFDs of the network traffic under different control strategies.


Average stopping time for different vehicle accumulations with different intersection control strategies. The left subfigure shows the average time of full stopping of the vehicles; the right shows the average time when the vehicle speed is less than 2m/s, i.e., slow-moving.


The intersection throughput and network-wide outflow under different control strategies. The throughput in the upper subfigure is the average number of vehicles passing through a single intersection per minute, and the outflow in the lower subfigure is the average traffic flow at the network exit links.


The accumulation of vehicles in the road network and the intersection control zone for different vehicle arrival rates, where the latter is the average vehicle accumulation at the four intersections within the network.


A snapshot of the simulation platform under FIFO based cooperative driving strategy.


The average travel time of vehicles in two typical routes, where travel time is defined as the time of vehicles from entering to leaving the boundary of the considered road network. The length of the red route is about 900 m, and the length of the green route is about 1000 m.


The MFDs of network traffic with different car-following headway settings. The left subfigure corresponds to the MCTS based cooperative driving strategy, and the right subfigure corresponds to the FIFO based cooperative driving strategy. The FIFO baseline in the left subfigure is the average of the four FIFO curves in the right subfigure, and the MCTS baseline in the right subfigure is the average of the four MCTS curves in the left subfigure.


The statistical zone used to count the gap between successive vehicles in the road network, where the length of the zone is 200 m.


The distribution of the gap distance under different vehicle accumulation.


Two typical heterogeneous traffic demand patterns. The traffic demand of the entrances indicated by the red arrows will be two or three times that of the entrances indicated by green arrows.


MFDs of network traffic under different turning ratios. Turning #Pattern A in the left subfigure: the proportion of left-turning vehicles is 1/3, straight-traveling is 1/3, and the right-turning is 1/3. Turning #Pattern B in the right subfigure: the proportion of left-turning vehicles is 1/4, straight-traveling is 1/2, and right-turning is 1/4.


A typical urban traffic network with twenty-five intersections. All straight links are 500 m long.


MFDs of the traffic network with twenty-five intersections under different control strategies.


仿真视频

Signal-free road network traffic

Citation

If you find our work is useful in your research, please consider citing:

@article{ZHANG2022103503,
  title = {Analysis of cooperative driving strategies at road network level with macroscopic fundamental diagram},
  journal = {Transportation Research Part C: Emerging Technologies},
  volume = {135},
  pages = {103503},
  year = {2022},
  issn = {0968-090X},
  doi = {https://doi.org/10.1016/j.trc.2021.103503},
  url = {https://www.sciencedirect.com/science/article/pii/S0968090X21004897},
  author = {Jiawei Zhang and Huaxin Pei and Xuegang (Jeff) Ban and Li Li},
  keywords = {Cooperative driving, Network traffic, Passing order, Macroscopic fundamental diagram (MFD)}
}
张嘉玮
张嘉玮
清华大学博士研究生

我的研究兴趣包括自动驾驶,智能汽车,智能交通系统,深度强化学习.