车路协同环境下道路交通安全研究进展

Research progress of road traffic safety in cooperative vehicle infrastructure environment

作为智慧交通发展的重要内容,车路协同技术通过人-车-路-环境之间实时通信主动感知交通安全风险,对预防道路交通事故具有重要作用。为了探究基于车路协同技术的道路交通安全研究进展,从车辆紧急碰撞与危险预警、车辆防追尾控制、不良驾驶行为分析、交叉口冲突分析及道路安全风险评估等方面概述了相关研究成果,梳理了相关研究方法、理论模型及系统架构,发现当前研究重点围绕微观驾驶行为建模和仿真评估、基于虚拟现实技术的驾驶模拟实践、多层次控制方法融合的评估与优化、安全场论视角下的道路风险评估等开展。通过梳理发现以下不足:首先,当前的碰撞模型构建过程存在一定限制,未综合考虑人-车-路-环境等关键特征要素,尤其基于驾驶人多维特征要素(如生理、心理、行为等)的碰撞风险建模不足,因此所建模型往往夸大或者低估实际风险水平,与此同时,多源信息获取与融合研究未充分考虑各种传感器和数据源的多样性,忽略了数据关联和权重分配的误差问题,尤其基于车载和路侧传感器融合的大范围道路交通信息感知研究亟待加强;其次,由于车路协同环境下驾驶人容易对自动驾驶系统产生过度依赖,从而丧失危险判断能力,目前研究缺乏对驾驶人判断决策变化机制、自动驾驶系统与驾驶人多模态交互机理、驾驶人安全接管可靠性评价的深入探讨;接着,对不良驾驶行为的研究尚缺少密集交通流条件下车辆集群间危险驾驶行为的量化分析,对车辆集群之间的交互作用和协同行为的研究亟待关注;最后,在交通冲突分析与安全风险评估方面,现有研究未探索一定时空范围多类型交通风险诱发与转化机理,基于多源交通风险叠加的道路交通安全量化评估仍是亟待解决的研究难题。综上,从感知融合、效能评价、集群分析、叠加量化等角度进行了未来展望,为车路协同环境下道路交通安全研究及技术应用提供参考。

As a major component of intelligent transportation systems,cooperative vehicle infrastructure technology actively identifies traffic safety risks through real-time communication between vehicles,humans,roads,and the environment and thus plays a critical role in preventing road accidents.To explore the research progress of road traffic safety based on vehicle road coordination technology,this study summarizes the relevant research results in the fields of vehicle emergency collision and risk warning,vehicle anti-rear-end control,bad driving behavior analysis,intersection conflict analysis,and road safety risk assessment,and combs the relevant research methods,theoretical models,and system architectures.Current research has focused on microscopic driving behavior modeling and simulation evaluation,driving simulation practice based on virtual reality technology,evaluation and optimization of multi-level control method fusion,and road risk assessment from the perspective of safety field theory.Through combing,the following deficiencies are found:First,some limitations exist with the current collision model construction process,and the key characteristic elements,such as the human-vehicle-road environment,have not been comprehensively considered.In particular,collision risk modeling based on multidimensional characteristic elements of the driver(such as physiology,psychology,and behavior)is insufficient.Therefore,the model often exaggerates or underestimates the actual risk levels.Simultaneously,multi-source information acquisition and fusion do not fully consider the diversity of various sensors and data sources and ignore errors related to data association and weight distribution.In particular,research on large-scale road traffic information perception based on vehicle and roadside sensor fusion must be strengthened.Second,due to the driver’s overreliance on automatic driving systems in a cooperative vehicle infrastructure environment,the driver loses the ability to judge dangers.To date,few studies have conducted in-depth investigations of the driver decision-making mechanism,multimode interaction mechanism between the automatic driving system and artificial driver,and reliability evaluation of the safety takeover of an artificial driver.Research on poor driving behavior still lacks a quantitative analysis of dangerous driving behavior between vehicle clusters under dense traffic flow conditions,and the interaction and cooperative behavior between vehicle clusters must be investigated.Finally,in terms of traffic conflict analysis and safety risk assessment,existing research has not explored the mechanism of multi-type traffic risk induction and transformation within a certain time and space range.The quantitative assessment of road traffic safety based on multisource traffic risk superposition remains an urgent research problem to be solved.This research was conducted from the perspectives of perception fusion,efficiency evaluation,cluster analysis,and superposition quantification,thus providing a reference for road traffic safety research and technology applications in cooperative vehicle infrastructure environments.