Zachry Department of Civil & Environmental Engineering, Texas A&M University, 3136 TAMU, College Station, TX, 77843-3136, United States.
Texas A&M Transportation Institute, 3500 NW Loop 410, San Antonio, TX, 78229, United States.
Accid Anal Prev. 2021 Jul;157:106162. doi: 10.1016/j.aap.2021.106162. Epub 2021 May 11.
The engagement of secondary tasks, like using a phone or talking to passengers while driving, could introduce considerable risks to driving safety. This study utilizes a near-crash dataset extracted from a naturalistic driving study to explore the patterns of near-crash events with or without the involvement of secondary tasks as a surrogate approach to understand the impact of these behaviors on traffic safety. The dataset contains information about driver behaviors, such as secondary tasks, vehicle maneuvers, other conflict vehicles' maneuvers before and during near-crash events, and the driving environment. The patterns for near-crashes with or without the involvement of secondary tasks are mined by adopting the apriori association rule algorithm. Finally, the mined rules for the near-crash events with or without the involvement of the secondary tasks are analyzed and compared. The results demonstrate that near-crashes with the involvement of secondary tasks often occur with drivers in a relatively stable and presumably predictable environment, such as an interstate highway with a constant speed. This type of near-crash is highly associated with the leading vehicle's sudden slowing or stopping since there is no expectation of any interruptions for these drivers performing the secondary tasks. The most common evasive maneuver in this kind of emergency is braking. Near-crashes without the involvement of secondary tasks is often associated with lane-changing behavior and sideswipe incidents. With shorter reaction time and awareness of the driving environment, the drivers in this type of near-crash can often make more complex maneuvers, like braking and steering, to avoid a collision. Understanding the patterns of these two types of near-crash incidents could help safety researchers, traffic engineers, and even vehicle designers/engineers develop countermeasures for minimizing potential collisions caused by secondary tasks or improper lane changing behaviors.
驾驶员在驾驶时从事次要任务，例如使用手机或与乘客交谈，可能会给驾驶安全带来相当大的风险。本研究利用从自然驾驶研究中提取的近碰撞数据集，通过探索有或没有次要任务参与的近碰撞事件模式，作为一种替代方法来了解这些行为对交通安全的影响。该数据集包含有关驾驶员行为的信息，例如次要任务、车辆操纵、近碰撞事件前后其他冲突车辆的操纵以及驾驶环境。通过采用先验关联规则算法挖掘有无次要任务参与的近碰撞模式。最后，分析和比较了有无次要任务参与的近碰撞事件的挖掘规则。结果表明，有次要任务参与的近碰撞通常发生在驾驶员处于相对稳定且可预测的环境中，例如限速的州际公路。这种类型的近碰撞与前车突然减速或停车高度相关，因为驾驶员在执行次要任务时没有任何中断的预期。在这种紧急情况下最常见的避让操作是制动。没有次要任务参与的近碰撞通常与变道行为和刮擦事件有关。由于反应时间较短且对驾驶环境有一定的认识，这种类型的近碰撞中的驾驶员通常可以进行更复杂的操纵，例如制动和转向，以避免碰撞。了解这两种类型的近碰撞事件的模式可以帮助安全研究人员、交通工程师，甚至车辆设计师/工程师制定对策，以最小化次要任务或不当变道行为引起的潜在碰撞的可能性。
