VRUs Safety Models Based on Risk Exposure for Safe Routing Algorithms

Current routing applications allow for the selection of routes based on factors like cost, travel time, or distance but do not consider safety concerns with regard to different types of travel modes and travel times. This poses an issue for the safety of vulnerable road users (VRUs). To create safe route options, it is crucial to assess accident probabilities at various sections and intersections. Measuring only the accident frequency while ignoring the time-dependent traffic count at locations is insufficient, as it does not allow to assess a persons’ accident risk. Additionally, accident frequencies underly statistical fluctuations which requires the implementation of accident models. Another important consideration for VRUs is their sensitivity to detours. Safe route options might deviate from the shortest path, so a spatial weighting method is needed to balance safety benefits against longer or complex routes. In the EU-funded project “SOTERIA” statistical models as well as machine learning techniques are being used to assess road design variants, highlight accident-prone areas, and identify factors leading to accidents. These models will guide the creation of safe routes to provide users with information on possible safety benefits.