What gaze behavior do pedestrians take in interactions when they do not understand the intention of an automated vehicle?

This study focuses on establishing an external human-machine interface (eHMI) for automated vehicles (AVs) that can clearly and quickly convey driving intentions to pedestrians, thus improving the acceptability of AVs. Frist of all, this study seeks to evaluate whether pedestrians clearly receive the information from the AV. This paper proposes a hypothesis based on a decision-making model of the pedestrian. If a pedestrian does not accurately understand the driving intentions of AVs, then his/her gaze duration at AVs will increase. A pedestrian--vehicle interaction experiment was designed to verify the proposed hypothesis. An AV was used for interacting with pedestrians but whether the AV stops or not during interactions that was controlled by the experimenter. The gaze data of pedestrians and their subjective evaluations of the AV's driving intentions were observed. The experimental results supported the hypothesis, i.e., there was a correlation between the participants' gaze duration at an AV and their understanding of the AV's driving intention. Moreover, the gaze duration of most participants at a manually driving vehicle was shorter than that at an AV. Besides, we proposed two recommendations to the designers of eHMI: (1) when a pedestrian is engaged in an interaction with the AV, the driving intentions of the AV should be provided; (2) if the pedestrian still gazes at the AV after the AV displays its driving intentions, the AV should provide more clear information about its driving intentions.