Interaction between automated vehicles and pedestrians in scenarios with enhanced uncertainty

Abstract

Autonomous driving has been getting a lot of attention in research in the last years. To enable a safe integration into a mixed traffic environment, autonomous vehicles (AVs) need to communicate with other, more vulnerable road users. Previous research has therefore proposed external Human Machine Interfaces (eHMI) and positive effects have been shown by current studies. This thesis aims at evaluating existing eHMI designs in form of a 360° light band in situations with multiple attendant interaction partners, since studies including such situations have been seldom so far. The main research question is how the perception-based eHMI design is transmittable to situations with enhanced uncertainty and whether the additional pedestrian brings up any negative effects. A video online study was conducted (N = 61), focusing on the participants' willingness to cross, feeling of being addressed, subjective certainty in their crossing decision and the amount of influence of the eHMI on the crossing decision when interacting with the different eHMIs. Furthermore, the effects of an additional pedestrian and its position on the said variables were investigated. The videos varied in the displayed eHMI design, the presence and positioning of an additional pedestrian and, in the perception-based eHMI, who and how many pedestrians were detected by the AV. Results showed significant differences between the eHMI designs but not for the presence and position of the additional interaction partner. Based on the results, it was concluded that no negative effects arose from the other attendant pedestrian and that the clear addressing and the interpretation of the perception-based eHMI could be maintained in scenarios with an additional pedestrian.