Collaborative Research: New Methods for Measuring, Evaluating and Predicting the Safety Impact of Road Infrastructure Systems on Driver Behavior

在这个合作研究项目中，研究人员研究了道路基础设施对驾驶员行为的影响及其对车辆与车辆相互作用的影响，以及对评估宏观交通网络性能的影响。针对高速公路和高速公路的几何特征，试图从交通流理论的角度理解道路周围环境对攻击性驾驶行为的影响，从而更好地理解几何对交通运营的影响、几何对安全的影响以及运营对安全的影响。为此，研究人员采用了统一的交通流框架，其中发展的微观加速模型和宏观双流体模型基于风险下的效用最大化(选择)。在了解非交通相关刺激的影响的同时，在相应的模型中明确地纳入风险态度和感知参数，可以量化不同道路路段对驾驶员行为的安全影响。为了校准基于道路特征的解释风险态度和感知的交通流模型，进行了广泛的数据收集。调查人员关注三个地理上不同的地点：华盛顿特区、大都市区以及洛杉矶和加利福尼亚州。利用交通事故数据以及数据采集点的公路基础设施特征，采用结构方程建模方法对新方法进行了验证。不同的研究结果被用来开发高速公路和高速公路上的替代安全措施，并创建一个原型微观仿真模型，捕捉外部非交通相关特征对不同模型参数的影响。从更广泛的角度来看，交通问题的特征是找到最有效的方法，以快速、安全的方式将人/货物从始发地运送到目的地。在车辆交通中，这些问题转化为避免道路拥堵，减少交通事故造成的人员和物质损失。这些问题的主要“参与者”是驾驶员/车辆单位(驾驶员的社会人口特征、车辆特性等)和这些单位周围的驾驶环境(环境条件、交通状况、交通基础设施系统特征等)。在过去的几年里，了解民用基础设施系统对司机行为的影响，特别是相应的安全影响，一直局限于识别“黑点”，并定性或统计地评估可以采取的安全措施，以避免或应对此类事件情景。如今，随着车辆行驶里程数的增加和相关事故的发生，在这一领域进行实质性研究的必要性同样重要，但更普遍的模型有助于回答重要问题：受道路几何特征影响的驾驶员的认知特性是什么？在这种影响中，哪些社会人口驱动因素特征和环境/天气特征起主要作用？司机之间有什么关系？行为和集体交通模式，包括拥堵动态？如何将与天气相关和与基础设施相关的参数纳入交通流量模型？需要回答这些基本问题，以便为更安全和高效的运输系统制定良好的基础设施设计战略和工程解决方案。通过改进交通管理控制系统和考虑冒险行为，所得到的交通流模型在与极端和危险条件相关的风险影响驾驶行为的疏散管理中具有巨大的价值。

In this collaborative research project, the investigators study the impact of roadway infrastructure on driver behavior and its implications on vehicle to vehicle interactions as well as on assessing macroscopic transportation network performance. Focusing on the geometric characteristics of highways and freeways and trying to understand how the road surrounding environment affect the aggressive driving behavior from a traffic flow theory perspective, better insight is given on the geometric effects on traffic operations, geometric effects on safety and effects of operations on safety. For that, the investigators adopt a unified traffic flow framework where the developed microscopic acceleration model and the macroscopic two-fluid model are based on utility maximization (choice) under risk. Explicit incorporation of risk attitudes and perception parameters in the corresponding models while understanding the influence of non-traffic related stimulus allows quantifying the safety implications of different roadway sections on driver behavior. In order to calibrate the traffic flow models that explain risk attitudes and perception based on road features, extensive data collection is undertaken. The investigators focus on three geographically diverse locations: Washington D.C. metropolitan area and the states of LA and CA. Crash data as well as highway infrastructure characteristics of the data collection sites are used for verifying the new approaches with the SEM (Structural Equation Method) modeling approach. The different research findings are used to develop surrogate safety measures on freeways and highways and to create a prototype microscopic simulation model capturing the impact of external non-traffic related characteristics on the different model parameters.From a broader perspective, transportation problems are characterized by finding the most efficient methods to move people/goods from an origin to a destination in a fast and safe manner. In vehicular traffic, these problems translate into avoiding road congestion and reducing human and physical losses due to traffic incidents. The major "players" in these problems are the driver/vehicle units (driver socio-demographic characteristics, vehicle properties, etc) and the driving environment (environmental conditions, traffic conditions, transportation infrastructure system characteristics, etc) surrounding such units. In the past years, understanding the impact of civil infrastructure systems on driver behavior, in particular the corresponding safety implications, have been limited to identifying "black-spots" and assessing qualitatively or statistically the safety measures that can be taken to avoid or respond to such incident scenarios. Nowadays, with the rising number of vehicles-miles travelled and the associated incidents, the need for substantial research in this area is no less important but with more generalized models helping answering important questions: what are the driver cognitive properties that are influenced by the road geometric features? Which socio-demographic driver characteristics and environmental/weather features play the major roles in such influence? What are the relationships between drivers? behavior and collective traffic patterns including congestion dynamics? How can weather-related and infrastructure-related parameters be incorporated in traffic flow models? These fundamental questions need to be answered in order to develop good infrastructure design strategies and engineering solutions for safer and efficient transportation systems. By improving traffic management control systems and considering risk-taking behavior, the resulting traffic flow models have tremendous value in evacuation management where risk associated to extreme and hazardous conditions affect driving behavior.