Do high visibility crosswalks improve pedestrian safety? A correlated grouped random parameters approach using naturalistic driving study data

Highlights

• Effectiveness of various High Visibility Crosswalk (HVC) types was evaluated.

• Naturalistic Driving Study data were used to identify driving behavioral change.

• Various safety surrogate measures were statistically modeled.

• Correlated grouped random parameters models with panel effects were estimated.

• The HVC presence was found to modify driving behavior and improve pedestrian safety.

Abstract

In this study, the effectiveness of High-Visibility Crosswalks (HVCs) in improving pedestrian safety at urban settings is assessed using SHRP2 (Second Strategic Highway Research Program) Naturalistic Driving Study (NDS) data. Various HVCs located at different positions on the roadway segment (mid-block vs end-of-block) and featuring different HVC marking designs (continental, bar-pair, and ladder) were selected for the assessment. As no pedestrian-vehicle crashes or conflicts were identified from the forward-facing videos and time series information of the SHRP2 Naturalistic Driving Study data, crash surrogate measures (i.e., speed; acceleration; throttle pedal actuation; and brake application) were employed to identify and analyze modifications in driving behavior at or near the HVCs.

The surrogate measures were statistically modeled using a correlated grouped random parameters estimation framework. This can account for panel effects arising from multiple traversals undertaken by each participant, for the effect of unobserved characteristics, as well as for their unobserved correlations, which constitute possible misspecification issues of statistical modeling. The results of the analysis showed that the presence of HVC modifies driving behavior, thus reducing the risk of motor vehicle – pedestrian conflicts. Apart from the presence of HVC, the HVC type (ladder, continental or bar-pair), the HVC location (mid-block or end-of-block) and various driver, roadway and trip characteristics were found to affect the vehicle speed, acceleration, throttle pedal actuation, and brake application.

Introduction

Pedestrians have been long identified as one of the most vulnerable groups of roadway users. Due to their significant physical exposure to roadway hazards, pedestrian-involved accidents are more likely to result in serious or fatal injuries compared to any other motorist-involved accidents. Interestingly, in 2016, a pedestrian casualty was observed approximately every 1.5 hours in road accidents in the United States (NHTSA, 2016). Previous studies of pedestrian-involved crashes and conflicts (Zegeer et al., 2005, Papadimitriou et al., 2009, Mitman et al., 2010, Paleti et al., 2010, Aziz et al., 2013, Haleem et al., 2015, Behnood and Mannering, 2016, Alhajyaseen and Iryo-Asano, 2017, Xin et al., 2017, Stapleton et al., 2017, Wang et al., 2019) have shown that drivers’ failure to yield to pedestrians constitutes a major contributing factor of pedestrian-vehicle crashes. Overall, the level of pedestrian safety has been found to be determined by vehicles’ speed and driver’s reaction time, especially in urban settings (Tefft, 2013, Jurecki and Stańczyk, 2014, Yasmin et al., 2014).

High Visibility Crosswalks (HVC) constitute one of the pedestrian safety countermeasures aiming to increase the upstream visibility of the crosswalks to the drivers as well as drivers’ consciousness about the possible presence of pedestrians. HVCs consist of pavement marking patterns (e.g., transverse lines, solid markings, ladder or continental markings, and so on) that can be easily detected from longer distances. Despite their low cost and ease of installation, the pedestrian hazards that are tackled by this type of crosswalk pavement markings have not been fully outlined to date. Previous studies were mostly devoted to the effectiveness of various types of HVCs in eliminating motor vehicles – pedestrian conflicts using either observed field data or data from driving simulation experiments (Gómez et al., 2013, Samuel et al., 2013). However, these studies did not capture drivers’ behavioral responses to the presence of crosswalks, or they did not control for the interaction of such responses with the prevailing weather, roadway, vehicle or traffic conditions. To account for the effect of the aforementioned conditions in the context of a comprehensive safety appraisal of HVCs, the naturalistic driving study (NDS) data from the second Strategic Highway Research Program (SHRP2) are used. The latter can provide a wide spectrum of drivers’ behavioral nuances, along with high-dimensional vehicle and road environment information (Campbell, 2012, Hamzeie et al., 2017). For the assessment of HVCs, safety surrogates (i.e., speed, acceleration, throttle pedal actuation, and brake pedal state) are employed (Hadi and Thakkar, 2003, Guo et al., 2010, Tarko et al., 2011, Mohamed and Saunier, 2013, Wang and Stamatiadis, 2014, Vedagiri and Killi, 2015, Dougald, 2016, Sarwar et al., 2017b, Pantangi et al., 2020) as no pedestrian-vehicle crashes or conflict incidents were identified in the vicinity of crosswalks during the study period.

This study builds upon a previous, preliminary analysis of the effectiveness of HVCs at uncontrolled locations carried out by Sarwar et al. (2017a); in that study, three ladder-style high visibility crosswalks at uncontrolled locations in the Erie County, NY were investigated in terms of their effectiveness to decrease the occurrence and intensity of crash surrogates. In this work, we provide a more comprehensive assessment of the HVC effectiveness by focusing on various HVCs across multiple States, on different in-block locations, and on different crosswalk configuration types. For this purpose, SHRP2 NDS data corresponding to an extensive set of HVC traversals were obtained and statistically analyzed to understand whether and how the presence of HVCs amends drivers’ behavioral patterns.