Design and Evaluation of a Connected Work Zone Hazard Detection and Communication System for Connected and Automated Vehicles (CAVs) (03-050)

Project Description: Roadside work zones present imminent safety hazards for roadway workers as well as passing motorists. The advent of connected and automated vehicles (CAVs) are driving work zone safety practitioners and vehicle designers towards implementing solutions that will more accurately describe activity in work zones to help identify and communicate imminent safety hazards that elevate crash risks. The aim of this project is to deliver a real-time threat detection and warning algorithm that is able to accurately localize, monitor, and predict work zone actors’ collision threat based on their movements and activities. This information along with CAV’s trajectories was used to detect potential proximity conflicts and provide advanced warnings to workers, passing drivers, and automated vehicle control systems. To this end, data regarding various roadway construction activities and movements were collected through four data collection sessions. The worker and construction equipment data was collected using handheld Ultra-WideBand (UWB) sensors, and the respective CAV data was received from On Board Units(OBU). The data includes location information such as latitude, longitude for workers and latitude, longitude, heading, speed and vehicle dimensions. The data was used to train the machine learning algorithm developed to recognize worker activities. As a result, this research provides a key element required to significantly improve the safety conditions of roadside work zones through prompt detection and communication of hazardous situations to workers and CAVs alike. Data Scope: A total of 82626 datapoints was collected from UWB tags and 129428 was collected from CAVs during four data collection sessions at Virginia Smart Roads. The data was collected in the Highway section in two settings: straight road and curve road. This “ConnectedWorkzone” dataset includes the 58 different variables. Data Specification: The ConnectedWorkzone dataset contains the collected data and the annotation regarding each data type. The UWB data includes tag IDs, time stamp, latitude and longitude of anchors (stationary) and , latitude and longitude of 4-6 tags(moving) as well as converted Cartesian coordinates. The CAV data includes RSEID, time stamp, latitude, longitude, speed, heading, dimensions, ABS, steering angle and acceleration.

项目说明：道路施工区域对道路作业人员及过往机动车驾驶人均存在紧迫的安全隐患。联网自动驾驶汽车（Connected and Automated Vehicles, CAVs）的出现，正推动施工区域安全从业者与车辆设计师研发更精准描述施工区域动态的解决方案，以识别并通报会提升碰撞风险的紧迫安全隐患。本项目旨在开发一套实时威胁检测与预警算法，可基于施工区域各类主体的运动与行为，精准定位、监测并预测其碰撞威胁。该信息结合联网自动驾驶汽车的行驶轨迹，可用于检测潜在的近距离冲突，并向作业人员、过往驾驶员及自动驾驶车辆控制系统提前发出预警。为此，本研究通过四次数据采集流程，收集了各类道路施工作业活动与运动相关的数据。作业人员与施工设备的数据通过手持式超宽带（Ultra-WideBand, UWB）传感器采集，联网自动驾驶汽车的对应数据则从车载单元（On Board Units, OBU）获取。所采集的数据包含位置信息：作业人员的纬度、经度，以及车辆的纬度、经度、航向、速度与外形尺寸参数。上述数据被用于训练识别作业人员行为的机器学习算法。最终，本研究通过及时检测危险场景并向作业人员与联网自动驾驶汽车通报此类场景，为显著改善道路施工区域的安全状况提供了关键支撑。

数据范围：在弗吉尼亚智能道路（Virginia Smart Roads）开展的四次数据采集中，共从超宽带标签采集到82626条数据，从联网自动驾驶汽车采集到129428条数据。数据采集路段为高速公路，包含两种道路场景：直线路段与曲线路段。本"ConnectedWorkzone"数据集共包含58种不同变量。

数据规格：ConnectedWorkzone数据集包含采集所得的原始数据与各数据类型的标注信息。超宽带数据包含标签ID、时间戳、固定锚点的经纬度、4至6个移动标签的经纬度，以及转换后的笛卡尔坐标系坐标。联网自动驾驶汽车数据包含路边设备标识（RSEID）、时间戳、纬度、经度、速度、航向、车辆尺寸、防抱死制动系统（Anti-lock Braking System, ABS）、转向角与加速度参数。