The effect of cognitive load on driver situational awareness under partially automated driving conditions

With the development of automated driving technology, level 2 (L2) automated systems have transformed the role of drivers, leading to changes in cognitive loads (including both physiological and psychological factors). Under partially automated conditions, drivers are prone to fatigue and disengagement, which reduces Situation Awareness (SA). The distribution of attention and the level of SA are critical to driving performance and safety. To investigate how SA varies under different cognitive loads during extended periods of automated driving, a high-fidelity driving simulator and eye-tracking technology were employed to collect data. The impacts of the automation level (manual vs. L2), cognitive channel (visual vs. auditory), and cognitive load (no load, 0-back, or 1-back) on drivers’ SA during extended driving were examined. In the simulation, a suburban road environment, including potential hazard scenarios and SA measurement tasks, was modeled. The results indicate that cognitive load influences drivers’ subjective cognitive loads and situational awareness scores. Overall, the cognitive load during L2 automated driving is lower than that during manual driving. As the driving time increases, drivers’ situational awareness tends to decrease. After 40 min of driving, a decrease in the subjective situational awareness score occurs. The distribution of drivers’ gaze points is influenced by automation, cognitive load, and their interaction. Gaze transition probability influenced by driving duration and cognitive load level, with the gaze concentration effect occurring after prolonged driving. Additionally, as the driving time increases, eye movement indicators such as the number of fixations on potential hazards and fixation entropy decrease, whereas the pupil coefficient of variation increases. This study reveals the relationship between cognitive load and SA, showing that visual metrics can effectively reflect drivers’ SA. These results provide valuable insights for designing driving cues that reflect drivers’ current state.