Drivers' Knowledge of Automated Driving System and Take-Over Request Performances

Abstract

This study investigated the effects of drivers' knowledge of a conditionally automated driving system on their performances in take-over requests from the system while conducting a non-driving-related task. A driving simulator experiment was conducted to answer this research question. Thirty drivers participated in the experiment, fifteen of whom had previously experienced take-over situations in the system ; the others had not experienced and were also instructed in a different way. Participants were asked to regain manual control and avoid colliding with a stopped vehicle when the take-over request occurred while conducting the non-driving-related task. In the take-over request situation, participants were able to know the situation only through visual and auditory displays in the vehicle because they could not perceive it from the frontal scene and vehicle movement. The results showed that participants' knowledge positively affected their performance in the take-over request situation and participants who had knowledge experienced less collisions than participants without knowledge. The results also showed that participants without knowledge tended to execute the non-driving-related task again although they had to react to the changes of traffic situation. Human factor issues in the conditionally automated driving system are discussed in terms of drivers' performances.