Modeling Cycling Behavior to Improve Bicyclists' Safety at Intersections - A Networking Perspective

Abstract

Road traffic is continuously increasing worldwide and Vulnerable Road Users (VRUs)such as bicyclists are ever more susceptible to injuries from crashes. Vehicular Ad-hoc Network (VANET)technologies in combination with Advanced Driver Assistance Systems (ADAS)are currently being evaluated to enhance road traffic safety. The same technology is now also considered for warning cyclists about imminent crashes. Prototypes for such VANET -enabled bicycles already exist, however, the primary problem here is verification and test due to immediate safety concerns for the participants. Computer simulation is considered a more feasible alternative; however, realistic behavior models of VRUs are not available in the required quality and quantity. We developed a virtual cycling environment to address this issue, allowing to study and record cyclists' behavior. We collected (and make available)traces from a variety of subjects cycling towards an intersection for three different safety critical scenarios. As proof of concept, we then used these traces in our vehicular networking simulation framework to assess the impact of a simple VANET solution on the cyclists' safety. Our results demonstrate the need for such an integrated framework for empirical studies as well as for simulation-based exploration of system configurations.