Identifying hand gestures for pedestrian-driver communication.

Abstract

Road traffic is largely defined by clear rules and laws. However, there are certain situations that are ambiguous and in which explicit communication between road users is needed to resolve such misjudgements or ambiguities. Especially hand gestures can be used to coordinate traffic by conveying different intentions, which refer to one's own behaviour or the change in the behaviour of others. This paper aims to identify such gestures focusing on the pedestrian-driver communication. The paper follows two approaches: Firstly, an experimental study (N = 20) was conducted analysing gestures in predetermined scenarios in a virtual and real environment. Secondly, traffic observations were conducted to analyse and compare natural occurring gestures with those identified in the study. As a result, 18 relevant pedestrian gestures were identified. The type of gestures and the expressiveness of the execution did not differ between the virtual or real study environment. From this, it can be concluded that the virtual environment is an appropriate study environment for identifying and analysing gestures. Furthermore, the type of gestures did not differ between the experimental study and the traffic observations, but the naturally occurring gestures were executed less expressively compared to those from the study. Finally, the paper presents a suitable approach on how pedestrian gestures can be studied and provides a foundation for further research on the communication between pedestrians and automated vehicles (AVs).