Experimental design for multi-modal take-over request for automated driving

Abstract

Despite recent advancement in autonomous vehicles, limitations of autonomous driving system still remain. When the autonomous driving system encounters failure, it hands over dynamic driving task to the driver. In the event of a failure of the autonomous driving system, effective take-over request should be issued according to driver’s status to guarantee safety and comfort of a driver. To devise the optimal alert modalities for take-over request that can minimize the risk of failure in the autonomous driving system, the following factors were drawn out for the experimental design: 4 scenarios for take-over request through the case analysis of cause of autonomous driving disengagement; and verifiable vehicle alert modalities of visual, auditory, and tactile aspects for take-over request through the literature search and market research. Then, we have established a virtual reality simulator for experiment environment that can effectively identify driver’s status based on human factors. This study will find out effective multi-modality for take-over request through an analysis of statistically significant differences according to driver’s status based on human factors by drawing the quantitative/qualitative data based on scenarios for take-over request and combinations of vehicle alert modalities of visual, auditory, and tactile aspects obtained from the results of previous studies analyzed in this experiment environment.