Search

The future use of automated vehicles involves a great deal of uncertainty regarding various aspects, including the interaction with vulnerable road users. This interaction has been studied only to a limited extent so far. As a first step in examining this interaction, there is a need for detecting relevant conflict points involving both types of road users (automated vehicles and vulnerable road users). A conflict point is defined as a location where several infrastructure-related elements as well as surrounding-related elements increase the potential of accident occurrence. These conflict points should reflect both safety issues concerning vulnerable road users and operational issues involving automated vehicles. We present a methodology for identifying such conflict points in order to simplify the selection of adequate instances to be used in studies involving automated vehicles. The presented methodology is mainly based on two pillars: a literature study on current accidents to identify safety problems in conflict points and the analysis of operational aspects concerning the use of automated vehicles. Based on these pillars, a scenario toolbox is presented, emphasizing significant parameters that should be accounted for when choosing relevant scenarios that result in conflict points. The developed scenario toolbox can facilitate the choice of conflict points in future studies concerning automated vehicles.

Directional counts at network nodes, in particular at urban intersections, are currently obtained during short time periods using either camera observations, or manual traffic counts or by using vehicle identification hardware, such as Bluetooth or ANPR. Since lanes are in many cases shared, continuous lane based detection does not provide directional counts for the majority of signalized urban intersections and roundabouts. The purpose of this research is to develop a method, which is purely based on the fusion of automatically recorded probe data and cross section counts at node entries and exits. This approach allows for continuous directional count analysis and thus the identification of patterns over different time spans with no need for an in-situ investment with devices or counting personnel.