Why so serious? – Comparing two traffic conflict techniques for assessing encounters in shared space

Abstract

In Germany, approximately 2.7 million crashes occurred in 2019. Especially vulnerable road users (VRU) have a high risk of being seriously injured or killed in traffic. Within the safe system approach, changes to the traffic infrastructure have been implemented to increase VRU safety. The creation of so-called shared spaces, in which all road users are encouraged to negotiate priority, is part of these efforts. Even though the concept has been known and applied for more than 40 years, comparatively little is known about interactions between different road users and methods to quantify interactions in shared spaces. The aim of this study is to investigate similarities and differences in quantifying the level of severity of encounters between pedestrians and motorised vehicles applying the Swedish traffic conflicts technique (STCT) and the pedestrian-vehicle conflicts analysis (PVCA). The STCT integrates the factors conflicting speed (CS) and time-to-accident (TA) to arrive at a severity level. In contrast, with four factors, the PVCA integrates more elements: time-to-collision (TTC, corresponding to TA), severity of evasive action, complexity of evasive action, and distance-to-collision (DTC). Trajectory and video data of a shared space were recorded using the Application Platform for Intelligent Mobile Units (AIM) in Ulm, Germany. 1364 interactions were randomly selected. Due to different exclusion criteria, such as interaction partners not being a car or pedestrian, missing values, and detection errors, 69 encounters were available for analyses. Using the PVCA, nine encounters were classified as critical and 60 as non-critical interactions. In contrast, computing the values based on the STCT, only three of the 69 encounters were categorised as critical. The results of a Spearman rank correlation did not show a significant correlation between the severity categories of the PVCA and severity levels of the STCT (r = 0.03, p = 0.78). An additional analysis of the encounters ranked as critical by the PVCA but as non-critical by the STCT showed that all six encounters had a large temporal distance (> 2 s) combined with very small spatial distance (< 5 m for vehicles and < 2.5 m for pedestrians). While the PVCA and STCT yielded similar results in most encounters, this could not be confirmed for all. Results indicate that spatial distance may contribute to the severity of encounters between pedestrians and vehicles in a shared space.