Temporal changes in secondary task engagement and system use during partially automated driving: a hybrid naturalistic study

Partially automated driving systems that can control the vehicle's acceleration and lateral position are becoming increasingly common. On-road and driving simulator research suggest that the shift from vehicle operator to system supervisor occurring in partial automation may lead to greater disengagement and driver distraction. The naturalistic driving approach where drivers are observed during everyday driving is often considered more methodologically sound, yet naturalistic studies on partial automation use are sparse. This study bridges this gap by adopting a hybrid naturalistic driving method to investigate behavioural adaptations including system use and secondary task engagement resulting from operating Tesla's, Volvo's, and Nissan's partially automated systems for 7 weeks. Three conditions were considered for each vehicle: manual driving, partial automation use, and experimental control. Manual and partial automation conditions refer to when participants were left free to decide when to drive the vehicle in manual or partially automated mode, respectively (six days/week). The experimental control condition took place when participants had to drive the vehicle in manual mode with no option of using partial automation (one day/week). Engagement in secondary tasks including handheld phone use, consuming food or drinks, and browsing the phone was recorded through the adoption of an automated video-based machine vision system. Results showed that: automation use remained stable over time; an uptick in secondary engagement was observed over time, but no differences were found between the three conditions; no differences in secondary task engagement or automation use were observed between the three systems. Our data partly contradicts the existing literature on the human factors of partial automation use. It also underscores the need for more naturalistic research investigating temporal fluctuations in potentially distracting behaviours during partial automated driving.