Sleep and take-over in automated driving

Abstract

Abstract The issue of driving performance after sleeping becomes highly interesting when introducing automated driving in Level 4 (SAE). It is known that fatigue as well as hypovigilance are severe risk factors for traffic accidents while an effective countermeasure is sleeping. In Level 4 automation sleeping while driving becomes a possible scenario, including the issues of transitions. Several seconds for take-over as known from distraction studies are not enough to adjust the seat, regain control and recover after sleeping. Furthermore, when analyzing human performance after sleeping, the phenomenon of sleep inertia may affect reaction time and quality of performance. Take-over time and the effect of sleep inertia are investigated in a driving simulator study with 44 participants. As independent variable, the time after waking-up until taking over is varied: The participants had to take-over the driving task one minute, seven minutes or 15 min after waking up. A control group completed the task without sleeping. Lane keeping, deviation of line during the lane change task, braking time and mental workload served as dependent variables. There were no significant differences in driving performance between the groups when looking at driving performance, safety aspects and mental workload, hence no effect of sleep inertia was measured here. Further studies should investigate individual performance of more outliers and apply even more controlled measures to understand better if and how sleep inertia can become a problem for driving and what countermeasures should be applied. The study also shows, that take-over time of one minute leads to successful, but hectic take-overs, whereas a take-over time of 15 min induces fatigue in some drivers. We conclude that powernaps have a good chance to become a safe reality in Level 4 automation with a suitable take-over time between one and seven minutes.