Feng Fred, Bao Shan, Sayer James R, Flannagan Carol, Manser Michael, Wunderlich Robert
University of Michigan Transportation Research Institute, 2901 Baxter Road, Ann Arbor, MI, 48109, USA.
University of Michigan Transportation Research Institute, 2901 Baxter Road, Ann Arbor, MI, 48109, USA.
Accid Anal Prev. 2017 Jul;104:125-136. doi: 10.1016/j.aap.2017.04.012. Epub 2017 May 10.
This paper investigated the characteristics of vehicle longitudinal jerk (change rate of acceleration with respect to time) by using vehicle sensor data from an existing naturalistic driving study. The main objective was to examine whether vehicle jerk contains useful information that could be potentially used to identify aggressive drivers. Initial investigation showed that there are unique characteristics of vehicle jerk in drivers' gas and brake pedal operations. Thus two jerk-based metrics were examined: (1) driver's frequency of using large positive jerk when pressing the gas pedal, and (2) driver's frequency of using large negative jerk when pressing the brake pedal. To validate the performance of the two metrics, drivers were firstly divided into an aggressive group and a normal group using three classification methods (1) traveling at excessive speed (speeding), (2) following too closely to a front vehicle (tailgating), and (3) their association with crashes or near-crashes in the dataset. The results show that those aggressive drivers defined using any of the three methods above were associated with significantly higher values of the two jerk-based metrics. Between the two metrics the frequency of using large negative jerk seems to have better performance in identifying aggressive drivers. A sensitivity analysis shows the findings were largely consistent with varying parameters in the analysis. The potential applications of this work include developing quantitative surrogate safety measures to identify aggressive drivers and aggressive driving, which could be potentially used to, for example, provide real-time or post-ride performance feedback to the drivers, or warn the surrounding drivers or vehicles using the connected vehicle technologies.
本文利用现有自然驾驶研究中的车辆传感器数据，研究了车辆纵向急动度（加速度相对于时间的变化率）的特征。主要目的是检验车辆急动度是否包含可用于识别激进驾驶员的有用信息。初步调查表明，驾驶员在踩油门和刹车踏板操作时，车辆急动度具有独特特征。因此，研究了两个基于急动度的指标：（1）驾驶员踩油门时使用大幅正向急动度的频率，以及（2）驾驶员踩刹车时使用大幅负向急动度的频率。为了验证这两个指标的性能，首先使用三种分类方法将驾驶员分为激进组和正常组：（1）超速行驶，（2）跟车过近（追尾），以及（3）他们与数据集中的碰撞或险些碰撞的关联。结果表明，使用上述任何一种方法定义的激进驾驶员，其两个基于急动度的指标值都显著更高。在这两个指标中，使用大幅负向急动度的频率在识别激进驾驶员方面似乎具有更好的性能。敏感性分析表明，在分析中改变参数时，研究结果基本一致。这项工作的潜在应用包括开发定量替代安全措施，以识别激进驾驶员和激进驾驶行为，例如可用于向驾驶员提供实时或行程后的性能反馈，或使用车联网技术向周围驾驶员或车辆发出警告。
