Impact of splitter-island on pedestrian safety at roundabout using surrogate safety measures: a comparative study

Abstract

This study aimed to investigate the effects of the splitter-island on pedestrian safety at roundabouts, considering the interaction between pedestrians and vehicles. Traffic accident statistics have traditionally been used to assess traffic safety. This method has severe limitations when used to investigate the impact of a particular feature of traffic facilities on safety at a microscopic level. Most previous research on surrogate safety measures (SSMs) had, on the other hand, focused on studying the safety of traffic operation conditions. The impact of a particular geometric feature of a roundabout on safety has so far received little attention in the literature. Therefore, we evaluated pedestrian safety in the presence and absence of physical splitter-islands by taking advantage of comparative statistical analysis (SA) and Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) methods. The Towa-Cho roundabout in Nagano prefecture in Japan was chosen for this study because it allows for five approaches, three with a splitter-island and two without. From 33-hr video data, a total of 343 and 183 potential pedestrian-vehicle conflicts were detected in three crosswalks with splitter-islands and two crosswalks without splitter-islands, respectively. The minimum time-to-collision (TTCmin), post-encroach-time (PET), maximum speed (MaxS), and maximum deceleration-to-safety (DTS) measures were monitored using particle tracking velocimetry (PTV) software for potential pedestrian-vehicle conflicts. It was found that geometric differences were significantly reflected by SSMs, and TTCmin was the most efficient among all. It was concluded that the splitter-island had a better safety performance than the roundabout approach.