Analysis of implicit communication of motorists and cyclists in intersection using video and trajectory data

Abstract

The interaction of automated vehicles with vulnerable road users is one of the greatest challenges in the development of automated driving functions. In order to improve efficiency and ensure the safety of mixed traffic, automated driving functions need to understand the intention of vulnerable road users, to adapt to their driving behavior, and to show its intention. However, this communication may occur in an implicit way, meaning they may communicate with vulnerable road users by using dynamic information, such as speed, distance, etc. Therefore, investigating patterns of implicit communication of human drivers with vulnerable road users is relevant for developing automated driving functions. The aim of this study is to identify the patterns of implicit communication of human drivers with vulnerable road users. For this purpose, the interaction between right-turning motorists and crossing cyclists was investigated at a traffic light controlled urban intersection. In the scenario, motorists and cyclists had a green signal at the same time, but cyclist had right-of-way. Using the Application Platform for Intelligent Mobility (AIM) Research Intersection, trajectory and video data were recorded at an intersection in Braunschweig, Germany. Data had been recorded for four weeks. Based on the criticality metric post encroachment time (PET) and quality of the recorded trajectory, 206 cases of interaction were selected for further analyses. According to the video annotation, when approaching the intersection, three common communication patterns were identified: (1) no yield, motorists, who should yield to cyclists, crossed the intersection first while forcing right-of-way; (2) active yield, motorists, who were in front of cyclists, gave the right-of-way; (3) passive yield, motorists, who were behind cyclists, had to give the right-of-way. The analysis of the trajectory data revealed different patterns of changes in time advantage in these three categories. Additionally, the communication patterns were evaluated with regard to frequency of occurrence, efficiency, and safety. The findings of this study may provide knowledge for the implementation of a communication strategy for automated driving functions, contributing to traffic efficiency as well as ensuring safety in the interaction with vulnerable road users.