Dou Jinzhen, Xu Chang, Wu Wenyu, Xue Chengqi, Chen Shanguang
School of Mechanical Engineering, Southeast University, Nanjing, China.
College of Mechanical Engineering, Donghua University, Shanghai, China.
Traffic Inj Prev. 2025;26(5):506-514. doi: 10.1080/15389588.2024.2427865. Epub 2025 Jan 10.
Attention forms the foundation for the formation of situation awareness. Low situation awareness can lead to driving performance decline, which can be dangerous in driving. The goal of this study is to investigate how different types of pre-takeover tasks, involving cognitive, visual and physical resources engagement, as well as individual attentional function, affect driver's attention restoration in conditionally automated driving.
A two-phase study was conducted. In phase one, a visual attentional task was employed to measure the attentional function of driver. In phase two, a driving simulator experiment was conducted, where participants experienced a typical sequence of automated driving, takeover and manual driving. Three pre-takeover tasks were designed to divert drivers' attentional resources, including a visual-cognitive task, a visual-physical task, and a monitoring task (control group). Eye-tracking metrics, including pupil and gaze behavior, along with driving behavior, were assessed as dependent variables.
The visual-cognitive task showed the highest percentage of pupil dilation and significantly increased participant's response time, but it also had a positive effect on subsequent attention restoration. Moreover, the attentional task scores were positively correlated with horizontal gaze scanning and negatively correlated with takeover response time.
Pre-takeover tasks with cognitive resource engagement proves to be superior for attention restoration in conditionally automated driving. The drivers with better attentional function are able to reduce recovering time. These findings make it possible to predict drivers' attentional state by identifying type of pre-takeover tasks in conditionally automated vehicles. Based on this, the attentive user interfaces could be adaptively adjusted to provide valuable cues, ensuring a safe transition.
注意力是态势感知形成的基础。态势感知能力低会导致驾驶性能下降,这在驾驶过程中可能是危险的。本研究的目的是调查不同类型的接管前任务,包括涉及认知、视觉和身体资源参与以及个体注意力功能的任务,如何影响有条件自动驾驶中驾驶员的注意力恢复。
进行了一项两阶段研究。在第一阶段,采用视觉注意力任务来测量驾驶员的注意力功能。在第二阶段,进行了驾驶模拟器实验,参与者体验了自动驾驶、接管和手动驾驶的典型序列。设计了三项接管前任务来转移驾驶员的注意力资源,包括视觉认知任务、视觉身体任务和监测任务(对照组)。眼动指标,包括瞳孔和注视行为,以及驾驶行为,被作为因变量进行评估。
视觉认知任务的瞳孔扩张百分比最高,显著增加了参与者的反应时间,但它对随后的注意力恢复也有积极影响。此外,注意力任务得分与水平注视扫描呈正相关,与接管反应时间呈负相关。
事实证明,涉及认知资源参与的接管前任务在有条件自动驾驶中对注意力恢复更具优势。注意力功能较好的驾驶员能够减少恢复时间。这些发现使得通过识别有条件自动驾驶车辆中的接管前任务类型来预测驾驶员的注意力状态成为可能。基于此,可以对注意力用户界面进行自适应调整,以提供有价值的提示,确保安全过渡。