Hand gestures as a form of communicating crossing intent from pedestrians to Automated Vehicles

Abstract

Communication between pedestrians and drivers partially relies on nonverbal communication methods such as eye-contact and gestures. With the transition from manually driven vehicles to automated vehicles (AVs), pedestrians could lose the ability to communicate their intention to the driver. This study investigated the use of hand gestures as a new form of communication from the pedestrian to the AV. Twenty-six participants participated in a Virtual Reality (VR) experiment, in which they wore an Oculus Rift to interacted with AVs in a virtual environment. The movement of the participants was recorded and visualized through the use of a Xsens Link motion tracking suit, which provided the research with data about the hand gesture usage. The main independent variable of this study was the permission for the participant to use hand gestures to try to make the AV yield. The hand gesture increased the probability of the AV stopping for the participant. The second independent variable was the response of the AV through a message on an external-Human Machine Interface (eHMI). The participants went through four different scenarios. Therefore, both one-way communication and two-way communication were investigated in the same experiment. The participants were given the freedom to decide if they wanted to use the hand gesture. Aside from the hand gesture, the participants were asked to perform a forward step at the moment they felt safe to cross the road in the virtual environment, without actual crossing. Alongside the gathered data on movement of the participants, the research also included data gathered from questionnaires in which the participants were asked about their feeling of safety, assurance of being seen by the AV, the effect that the lack of eye-contact had on their decision making, difficulty predicting the behaviour of the AV, and their trust in communication involving AVs and hand gestures. The research found that the participants used hand gestures to communicate crossing intent to the AV around 80% of the time. The ability to use hand gestures did not improve the feeling of safety significantly, and made it more difficult for the participants to predict the behaviour of the AV. The results of the subjective measurements did show positive results for the hand gesture in combination with responses from the AV by the eHMI, as well as for the eHMI alone. It is concluded that participants were willing to use the hand gesture, and that the hand gesture only increased the subjective feeling of safety if the AV responds to the hand gesture via an eHMI.