Cooperative ADAS for Increasing Cycling Traffic Safety in EU Project XCYCLE

Abstract

In urban traffic, cyclists are often exposed to high risks especially when being in interaction with motorized traffic participants. Objectives of EU funded project XCYCLE (http://www.xcycle-h2020.eu) are to improve and increase cyclists’ safety. Focusing on urban traffic, the project addresses the use case of an intersection with right turning motorized traffic (especially HGV’s) and privileged cyclists going straight on. One of the main projects activities lies in developing an infrastructural supported cooperative advanced driver assistance system (C-ADAS) with the following technical workflow: dynamic data (especially trajectories acquired by the detecting infrastructure) is processed in real-time to get an understanding about relevant aspects of the traffic situation. Risk related information is derived with respect to the given use case by predicting potentially critical encounters. The outputs are used for sending situation and risk related information to the given road users (esp. the motorized traffic participant) using wireless communication standards like ITS-G5. In addition to this system network, the authors currently discuss with the Braunschweig municipality possibilities for implementing visual road instrumentation for positively influencing gaze behavior of drivers of non-instrumented vehicles. The overall technical concept will be implemented, integrated, tested, and evaluated in XCYCLE. The implementation of the demonstrator is done based on test field AIM (Application Platform for Intelligent for Mobility, http://www.dlr.de/ts/aim). The full manuscript gives an overview of the project activities. It focuses on the algorithm development for detecting and predicting safety relevant traffic situations for the given use case taking into account all relevant external factors. The manuscript gives a profound insight into the works of modelling cycling and driving behavior as well as interaction patterns for the given traffic scene. In addition, approaches for setting up predictive elements in situation and risk assessment are discussed that allow a projection of the given scene into the future. These are crucial to enable the overall system to intervene before a safety-relevant event actually occurs. All aspects are shown and discussed based on real long-term trajectory data that is acquired by the detecting infrastructure, operating 24/7. This infrastructure is already built-up and in operation since October 2016.