The impact of level of automation and lighting conditions on drivers’ visual scanning patterns in an urban environment

Elevated crash risk is observed in manual driving after dark, partly because reduced visibility constrains drivers’ visual scanning patterns. L2 assisted driving technologies may help compensate for these increased attentional demands by reducing drivers’ physical operational demands, thereby freeing attentional resources that may support more extensive visual scanning. However, research on L2 has largely focused on daytime motorway scenarios, leaving it unclear how L2 may affect performance during after dark urban driving, where peripheral traffic places unique attentional demands on drivers under reduced visibility. This driving simulator study employed a 2 x 2 within-participants design to examine the effects of automation levels (manual driving, L2 assisted driving) and lighting conditions (daytime, after dark) on visual scanning patterns in urban scenarios. We also investigated whether visual attention towards different types of ambient traffic (vulnerable road users, parked vehicles) differed across conditions. Results showed that horizontal gaze dispersion did not differ across automation levels. However, percentage of glance time analysis showed that drivers spent more time monitoring the peripheral environments during L2 assisted driving than manual driving. Regardless of levels of automation, drivers allocated greater visual attention towards VRUs over parked vehicles, suggesting a consistent prioritisation of safety-relevant dynamic elements regardless of the attentional resources available to them. However, this pattern was only observed during daytime conditions, reflecting drivers’ reduced ability to distinguish between different elements of the driving scene under reduced visibility. L2 assisted driving also reduced vertical gaze dispersion relative to manual driving, though the reduction was less pronounced after dark, potentially because ambient traffic was only perceivable at closer distances under reduced visibility and therefore appeared lower in the visual field, requiring greater vertical scanning. These results imply that future driving assistance systems may benefit from incorporating night-vision camera systems to better support drivers’ visual attention in the dark.