Adaptations of drivers’ gaze behaviour during car-following as a function of vehicle automation, duration of use and prior driving experience

Abstract Vehicle automation progressively shifts the driver’s role from active operator to passive supervisor, requiring behavioural adaptation. Although a timely issue, little is known about the influence of previous manual driving experience and the duration of automation use on those adaptations. Gaze behaviour during a car-following task were investigated under four automation conditions: Non-Automated (NA), Automated Steering (AS), Highly Automated Driving (HAD), and Fully Automated driving (FA). The sequences of visual explorations across the different areas of interest of the driving scene (i.e., scan paths) were analysed over seven weeks of automation use and for both unlicenced and experienced drivers. The findings reveal that automation alters the distribution of scan paths compared to manual driving. Specifically, automation promotes scenery and supervision scan paths at the expense of driving-control ones, which are crucial for monitoring the road and vehicle speed control. This shift became noticeable once the driver completely disengaged from the physical control of the vehicle, this function being supported by automation. Additionally, driving experience significantly influences gaze behaviour for each automation solution. Novice drivers focus more on speed-monitoring scan paths, whereas experienced drivers display more balanced scan paths. No significant changes in scan paths were observed over time, suggesting that gaze strategies are highly context-dependent. Altogether, this study provides a better understanding of the effects of automation on the dynamics of drivers’ gaze behaviour.