自主交叉口综述:设计与管理  

A review of autonomous intersections:design and management

作　　者：蒋浩然 赵萤滢 罗孝羚 姚志洪[1,2,4] 蒋阳升 JIANG Haoran;ZHAO Yingying;LUO Xiaoling;YAO Zhihong;JIANG Yangsheng(School of Transportation and Logistics,Southwest Jiaotong University,Chengdu 611756,China;National Engineering Laboratory of Integrated Transportation Big Data Application Technology,Chengdu 611756,China;Postdoctoral Research Station,Chongqing City Investment Golden Card Information Industry(Group)Co.,Ltd.,Chongqing 401336,China;National United Engineering Laboratory of Integrated and Intelligent Transportation,Chengdu 611756,China)

机构地区：[1]西南交通大学,交通运输与物流学院,成都611756 [2]综合交通大数据应用技术国家工程实验室,成都611756 [3]重庆市城投金卡信息产业(集团)股份有限公司,博士后科研工作站,重庆401336 [4]综合交通运输智能化国家地方联合工程实验室,成都611756

出　　处：《交通运输工程与信息学报》2025年第1期36-58,共23页Journal of Transportation Engineering and Information

基　　金：国家自然科学基金项目(72471200);四川省科技计划项目(2024NSFSC0179);中国博士后科学基金面上资助项目(2024MD764029)。

摘　　要：【背景】自主交叉口是智能网联环境下的新型城市道路交叉口,通过车辆实时通信与协作取代交通信号灯,实现安全高效通行,被视为未来城市交通发展的重要方向。【目的】为了探究自主交叉口的研究进展,从系统设计与交通管理两方面对相关研究成果进行梳理和总结。【方法】首先,介绍了自主交叉口的空间布局及冲突构建方法。其次,归纳了研究中常用的简化假设,基于冲突构建方法与假设,总结了自主交叉口管理的系统架构与理论模型。此外,介绍了两类用于验证管理方法效果的评估手段。最后,提出了未来研究的潜在方向。【结果】自主交叉口的空间布局包括通信与规划区域、控制执行区域以及冲突区域,其中冲突区域可通过栅格法、冲突点法或冲突区域法进行构建。自主交叉口管理方法具有基于规则、优化、学习和博弈四种建模方式。为简化建模,研究常在车辆特性、通信特性、速度与车道管理以及环境条件等方面进行假设。除了仅面向智能网联汽车的基础性自主交叉口管理外,扩展性研究包括混合交通环境下的交叉口管理、行人过街管理、多交叉口协同控制以及路网中交叉口的优化布局。自主交叉口管理在提升交叉口通行效率、保障交通安全、改善驾驶舒适性以及促进交通系统可持续发展方面具有显著成效。未来研究可在提升模型精确性、促进管理方法融合、增强测试环境真实性等方面进一步突破。【应用】本综述可为自主交叉口的设计、管理及实际应用提供理论支撑与参考。[Background]Autonomous intersections are a new type of urban road intersection in intelligent networked environments.Autonomous intersections replace traffic signals using real-time communication and cooperation of vehicles to realize safe and efficient passage,and are regarded as a vital direction for future urban traffic development.[Objective]To explore the research progress of autonomous intersections,relevant research results are sorted out and summarized in terms of system design and traffic management.[Methods]First,the spatial layout and conflict construction methods of autonomous intersections are introduced.Second,common simplification assumptions made in the research are summarized.Based on the conflict construction methods and assumptions,the system architecture and theoretical model of autonomous intersection management are summarized.Additionally,two commonly used evaluation techniques are presented.Finally,potential directions for future research are proposed.[Results]The spatial layout of an autonomous intersection includes the communication and planning area,control execution area,and conflict area,in which the conflict area can be constructed using the grid method,conflict-point method,or conflict-region method.Autonomous intersection management methods have four modeling approaches:rule-based,optimization-based,learning-based,and game-based.To simplify modeling,studies often make assumptions regarding the vehicle characteristics,communication characteristics,speed and lane management,and environmental conditions.In addition to basic autonomous intersection management for connected and automated vehicles,extended studies include intersection management in mixed traffic environments,pedestrian crossing management,multi-intersection cooperative control,and intersection layout optimization in road networks.Autonomous intersection management has the potential to effectively improve traffic efficiency at intersections,ensuring traffic safety,enhancing driving comfort,and sustainable development of tr

关 键 词：智能交通 自主交叉口 文献综述 设计与管理 网联自动驾驶汽车 无信号交叉口 

分 类 号：U491.23[交通运输工程—交通运输规划与管理]