GOALI/Collaborative Research: Human Maintenance - A Prognostics Framework to Model Changes in Drivers' Safety Performance and Optimize Dispatching Policies

这个GOALI学术联络机会(GOALI)项目将调查纳入分析工具的机会，以便对卡车司机的安全表现进行建模，并随后优化调度政策。这项研究的动机是，在美国和世界各地，交通事故仍然是一个紧迫的公共安全问题。与疲劳有关的司机表现恶化是造成致命道路事故的主要因素，特别是在涉及商业半挂车卡车的道路事故中。卡车司机在一个复杂而动态的环境中工作，目前对各种因素如何与司机安全履行所需职责的能力相互作用还没有足够的了解。与此同时，大量数据要么由卡车运输和运输公司例行收集，要么可以在其他地方获得。这些数据包括路线和休息时间表细节、司机记录的小时数、交通和天气状况、与驾驶相关的结果等。该研究项目旨在了解司机表现的变化如何作为这些数据集所代表的驾驶条件的函数来发展，以及随后如何将这些信息用于实际决策。这项研究与一个教育计划相结合，该计划的基石是一个在线平台，该平台将允许向当前和未来的实践者传播研究成果。假设驱动结果的恶化可以使用可靠性理论中的累积老化模型来建模。驾驶员安全性能的变化(由于疲劳、注意力集中、嗜睡或冒险)将被建模为使用卡车活动数据而不是更具侵入性的驾驶舱内记录(脑电、心率监测器等)的参数或非参数的老化函数。接下来，该模型将被用于在路径和调度策略中考虑安全和驾驶员的性能。这里的重点将放在框架的数据驱动性质上，该框架能够随着更多数据的出现而不断改进和完善模型。如果成功，该项目将为卡车运输业的决策提供一个全面的框架。

This Grant Opportunity for Academic Liaison with Industry (GOALI) project will investigate opportunities for incorporating analytical tools for modeling truck drivers' safety performance and subsequent optimization of dispatching policies. This research is motivated by the fact that transportation incidents remain a pressing public safety issue in the United States and throughout the world. Fatigue-related deterioration of driver's performance is a major factor contributing to fatal road incidents, especially among those involving commercial semi-trailer trucks. Truck drivers operate in a complex and dynamic environment, and currently there is not enough understanding of how various factors interact with the driver's ability to safely perform the required duties. At the same time, large amounts of data are either routinely collected by trucking and transportation companies or are available elsewhere. These data include route and rest schedule details, hours logged by the drivers, traffic and weather conditions, driving-related outcomes, etc. The research project aims at understanding how changes in a driver's performance develop as a function of driving conditions represented by those datasets, and subsequently, how this information can be used in practical decision making. The research is integrated with an education plan whose cornerstone is an online platform that will allow for the dissemination of the research outcomes to current and future practitioners.It is posited that deterioration in driving outcomes can be modeled using cumulative aging models from reliability theory. The changes in in a driver's safety performance (due to e.g., fatigue, attentiveness, sleepiness, or risk-taking) will be modeled as an aging function either parametrically or non-parametrically using truck activity data rather than more invasive in-cabin recordings (electroencephalograms, heart rate monitors, etc.). Next, this model will be employed to incorporate safety and driver's performance considerations in routing and dispatching policies. The focus here will be on the data-driven nature of the framework enabling continuous improvement and refining of the models as more data becomes available. If successful, this project will provide a comprehensive framework for decision making in the trucking industry.