Quantifying Drowsy Driving

Project Description This dataset was created to support a project intended to characterize the relationship between drowsy driving and crash risk in the SHRP 2 NDS, identify driver-critical reasons for crashes and near-misses in drowsy driving events, and identify individual differences including, but not limited to, demographics and personality characteristics, which correlate with the incidence of drowsy driving. Data Request Scope: This dataset contains: 2,048 trips extracted from the InSight query https://insight.shrp2nds.us/query/index#/builder?queryID=39023 A total of 785 trips which consisted of 585 crash/near-crash events previously coded as "Drowsy, sleepy, asleep, and fatigued" plus 200 baseline events without this impairment A subset of Insight-available variables and time series variables Video coding results. The video coding was performed at the VTTI data reduction laboratories to identify drowsiness based on whole body/face movements. Specifically, VTTI personnel coded Observer Rating of Drowsiness (ORD) ratings for 781 1-minute long trip segments of the SHRP 2 NDS. Data Specification This dataset contains different types of data from the SHRP 2 dataset: For the Event Detail Table variables, please see Insight_Westat_Dictionary.pdf For the time series variables please see TimeSeries_Dictionary.pdf and ProcessedRadarDictionary.pdf The forward video view (H.264 encoded MP4) The variables from the ORD coding are: SCE-BL_Event_ID: Event_ID of the crash, near crash, or baseline that was used in the SHRP2 Event Detail table (Insight). ORD_Event_ID: Event_ID of the 1-minute epoch created for the ORD ratings. These events start 1 minute prior to the precipitating event (when possible) and end at the precipitating event time. AVEORDRATING: Preliminary average of all current ORD ratings. NUMRATINGS: The number of preliminary ORD ratings that exist for each event. Event_Type : Designates each event as coming from the crash, near crash, balanced-sample baseline, or additional baseline pool. SampleSource: Designates each event as coming from the events previously flagged as Drowsy or Not Drowsy. AnonymousParticipantID: A numeric identifier assigned to a participant randomly during in-processing and used in place of any other driver identity information. Notes: Explanations for missing ORD ratings.

项目描述 本数据集旨在支撑一项研究项目，该项目旨在刻画战略公路研究计划2号全国驾驶数据集（Strategic Highway Research Program 2 Nationwide Driving Study, SHRP 2 NDS）中疲劳驾驶（drowsy driving）与碰撞风险（crash risk）之间的关联关系，识别疲劳驾驶事件中引发碰撞及险情（near-misses）的关键驾驶人诱因，并厘清与疲劳驾驶发生率相关的个体差异——包括但不限于人口统计学特征（demographics）与人格特质（personality characteristics）。
数据请求范围：本数据集包含如下内容：
1. 从InSight查询平台（https://insight.shrp2nds.us/query/index#/builder?queryID=39023）提取的2048条行程；
2. 共计785条行程，其中包含585起此前被标注为“困倦、嗜睡、昏睡及疲劳”的碰撞/险情事件，以及200条无此类状态的基准行程；
3. InSight平台可用变量子集及时序变量（time series variables）；
4. 视频编码结果（video coding results）。
视频编码工作于VTTI数据缩减实验室完成，通过捕捉全身及面部动作识别驾驶人疲劳状态。具体而言，VTTI工作人员对SHRP 2 NDS的781段1分钟时长的行程片段开展了疲劳观察者评分（Observer Rating of Drowsiness, ORD）标注工作。
数据规范：本数据集包含源自SHRP 2数据集的多类型数据：
- 事件详情表变量的详细说明，请参阅Insight_Westat_Dictionary.pdf；
- 时序变量的详细说明，请参阅TimeSeries_Dictionary.pdf与ProcessedRadarDictionary.pdf；
- 前置视角视频（采用H.264编码的MP4格式，H.264 encoded MP4）；
- ORD编码相关变量如下：
1. SCE-BL_Event_ID：对应SHRP2事件详情表（Insight平台）中所使用的碰撞、险情或基准事件的事件ID（Event_ID）；
2. ORD_Event_ID：为开展ORD评分而创建的1分钟时段事件ID。此类时段通常（若可行）始于触发事件前1分钟，终止于触发事件时刻；
3. AVEORDRATING：当前所有ORD评分的初步平均值；
4. NUMRATINGS：单个事件对应的有效初步ORD评分数量；
5. Event_Type：将事件划分为碰撞事件、险情事件、平衡样本基准事件或额外基准池事件；
6. SampleSource：将事件归类为此前被标记为“疲劳”或“非疲劳”的事件；
7. AnonymousParticipantID：在驾驶人入组流程中随机分配的数字标识符，用于替代所有可识别驾驶人身份的其他信息。
备注：针对缺失ORD评分的补充说明。

搜集汇总

背景与挑战

背景概述

该数据集聚焦于疲劳驾驶研究，包含2048次行程数据，其中785次涉及疲劳驾驶相关事件，旨在分析疲劳驾驶与事故风险的关系及个体差异因素。数据集提供了详细的视频编码和观察者评级数据，为疲劳驾驶行为研究提供了重要资源。