Evaluation of three different interaction designs for an automatic steering intervention

Abstract

The study was designed to test three different interaction designs for an automatic steering intervention against a baseline. Forty participants participated in a driving simulator study with four experimental groups. Each group experienced an imminent rear-end collision situation. The thirty drivers of the three automation groups were supported by an automatic steering intervention, each group experiencing a different interaction design: (i) pure automatic intervention, (ii) directed haptic warning plus automatic intervention, (iii) undirected acoustic warning plus automatic intervention. Ten participants underwent the emergency situation without any intervention (baseline). Additionally, participants of the three automation groups experienced an automatic steering intervention in a false alarm situation to test for controllability. It was hypothesized that the directed haptic warning plus intervention would support the effectiveness and the controllability most successfully. The results show that all automatic steering interventions supported an evasion manoeuvre in the imminent collision situation but no further difference in performance was found between the three interaction designs. Subjective data revealed that not all drivers recognized the steering intervention. For the false alarm situation, some drivers could not override the false automatic intervention and could not stabilize the vehicle in the lane. Further research is needed to improve the effectiveness of automatic steering interventions in emergency situations and the controllability in false alarm situations.