100-Car Data

The 100-Car Naturalistic Driving Study was an instrumented vehicle study conducted in the Northern Virginia / Washington, D.C. area over a 2 year period. The primary purpose of the study was to collect large scale naturalistic driving data. To this end, the instrumentation was designed to be unobtrusive, study participants were given no special instructions, and experimenters were not present. Approximately 100 vehicles were instrumented with a suite of sensors including forward and rearward radar, lateral and longitudinal accelerometers, gyro, GPS, access to the vehicle CAN, and five channels of compressed digital video. Collection rates for the various sensors ranged from 1Hz to 10Hz. This collection effort resulted in approximately 2,000,000 vehicle miles and 43,000 hours of driving data. Detailed video analysis has been compiled for 68 crashes and 760 near crashes.

100车自然驾驶研究（100-Car Naturalistic Driving Study）是一项在北弗吉尼亚/华盛顿特区区域开展的仪器化车辆研究，实验周期为2年。本研究的核心目标为采集大规模自然驾驶数据。为此，研究团队将测试仪器设计为低侵入式形态，未向受试参与者提供任何特殊驾驶指令，且实验人员全程不介入数据采集过程。本次研究共为约100台车辆加装了整套传感设备，涵盖前后向雷达、横向与纵向加速度计、陀螺仪、全球定位系统（GPS）、车辆控制器局域网（CAN）接口，以及5路压缩数字视频通道。各类传感器的采样频率介于1Hz至10Hz之间。本次数据采集工作累计获得约200万车英里的行驶数据与4.3万小时的驾驶记录。研究团队已针对68起碰撞事故与760起险情完成了详细的视频分析工作。

搜集汇总

背景与挑战

背景概述

100-Car Data数据集是一个大规模的自然驾驶研究数据集，包含约100辆车的传感器数据（如雷达、加速度计、GPS等）和视频记录，总计约2,000,000车辆英里和43,000小时的驾驶数据，特别针对68起碰撞和760起接近碰撞事件进行了详细分析。该数据集的特点是数据采集自然、无干扰，适用于研究真实驾驶行为和交通安全。