Multimodal Takeover Request Displays for Semi-automated Vehicles: Focused on Spatiality and Lead Time.

To investigate the full potential of non-speech sounds, this study explored the effects of different multimodal takeover request displays in semi-automated vehicles. It used a mixed design - the visual and auditory notification lead time was within-subjects, whereas the auditory notification spatiality was between-subjects. The study was conducted in a motion-based driving simulator with 24 participants. All participants were engaged in four 9-min driving tasks in level-3 automated vehicle and simultaneously performed a non-driving related task (NDRT, online game). Each driving session contained three hazardous events with takeover request (in total 12 requests per user). The results showed that 3-s lead time evoked the fastest reaction time but caused high perceived workload and resulted in unsafe and non-comfortable maneuver. In terms of workload and maneuver, 7-s lead time showed better results than others. Auditory displays with directional information provided significantly better reaction times and reaction types. Subjective evaluation, on the other hand, did not show any significant differences between non-directional and directional displays. Additionally, the results showed that braking is a more common first reaction than steering, and that the NDRT did not influence the takeover request.