Pedestrians’ Mental Model Development after Initial Encounters with Automated Driving Systems

Abstract

Automated driving systems (ADS) in urban road traffic incorporate external human-machine interfaces (eHMI) as explicit means of interaction with pedestrians. To enable safe interactions, a pedestrian’s mental model must be consistent with the functions and limitations of the ADS. The aim of this research is to investigate developments in pedestrians’ mental models towards ADS displaying different levels of automation transparency (AT) via an eHMI. Thirty-seven participants were instructed about ADS and experienced 60 interactions with an ADS on three test days (longitudinal design) in a controlled field test environment. As between-subject variable the level of AT was manipulated (status information vs. status, perception and yielding intention information). As dependent variable the mental model was measured by means of a questionnaire. The study results show that a pedestrian’s mental model of an ADS does not develop within the three initial encounters but is initially influenced by the level of AT conveyed via an eHMI. Participants in the high AT group assessed the ADS’s ability to communicate its yielding and nonyielding intention as more applicable. When designing eHMIs and instructing pedestrians accordingly, misconceptions in the pedestrians’ mental model should be considered.