Doctor of Philosophy (PhD)
Civil and Environmental Engineering and Construction
First Committee Member
Second Committee Member
Third Committee Member
Fourth Committee Member
Jin O. Choi
Fifth Committee Member
Number of Pages
Cities and metropolitan areas are increasingly facilitating pedestrians’ movement by the provision of pedestrian walking facilities. As pedestrian traffic increases, the risk of crash involvement increases, especially at midblock locations, where pedestrians are exposed to unsafe interactions with vehicular traffic. To improve pedestrians’ safety at midblock locations, various countermeasures are provided, which include signalized crosswalks. Several studies have analyzed driver-pedestrian interactions, as well as pedestrian-infrastructure interactions at signalized midblock crosswalks. However, more in-depth studies are necessary, due to shortfalls of study assumptions, which have led to the application of improper statistical models, as seen in the literature. Improved models are crucial, as they can be used to evaluate the factors affecting the effectiveness of countermeasures at signalized midblock crosswalks. Moreover, there are several aspects of pedestrian-infrastructure interactions that have not been studied in the previous research. This study, therefore, attempts to improve the methodologies for analyzing driver-pedestrian-infrastructure interactions at signalized midblock crosswalks. Specifically, this study is aimed towards:
• Developing improved modeling methodology for the yielding compliance of drivers at signalized midblock crosswalks, which considers the time taken to yield right of way, and the transition states undergone during yielding.
• Analyzing the risks associated with driver-pedestrian interactions at signalized midblock crosswalks.
• Developing the framework for modeling the spatial and temporal crossing compliance of pedestrians at signalized midblock crosswalks.
• Evaluating the influence of various crosswalk features, such as signs and markings, traffic-related variables, and pedestrian related factors on the safe utilization of signalized midblock crosswalks; these include factors influencing drivers’ yielding compliance, pedestrians’ crossing compliance, and pedestrians’ utilization of pushbuttons.
The study data were collected from a total of twenty signalized midblock crosswalks located in the Las Vegas, Nevada metropolitan area. These crosswalks have varying geometric configurations, signalizations, traffic characteristics, and pedestrian flows. Five types of signalization; Circular Flashing Beacons (CFBs), Circular Rapid Flashing Beacons (CRFBs), Rectangular Rapid Flashing Beacons (RRFBs), Pedestrian Hybrid Beacons (PHBs), and Traffic Control Signals (TCSs) were studied in this research. The observational survey method was applied for data collection, whereby video cameras were used to collect driver-pedestrian interactions. The data extraction was performed by reviewing the videos and recording the information of interest in a spreadsheet, with a total of 2638 pedestrians crossing incidents recorded for analysis. A descriptive analysis was performed, and several statistical models were developed.
Multistate hazard-based models are developed to model the yielding compliance of drivers. The transitional states while drivers are yielding right of way to pedestrians are defined as non-yield, “partial-yield” events (partial-yield, scenarios in which driver(s) in one lane yield, while other driver(s) in adjacent lane(s) in the same direction do not), and full-yield. Binary-based models are developed for modeling drivers’ spatial yielding compliance, pedestrians’ spatial crossing compliance, and pedestrians’ temporal crossing compliance. Rare Events Logistic Regression (RELR) is applied to evaluate the occurrence of partial-yield events and near-miss events. In addition to binary models, ordered models and multinomial models are developed and compared to model pedestrians’ spatiotemporal crossing compliance.
The results of the multistate models reveal that signal type, number of vehicles within effective crosswalk distance, yield-here sign, and crossing zone factors have similar influence for transition from non-yield to full-yield, non-yield to partial yield, and partial yield to full yield. Thus, the results of the binary models for yielding compliance are only partially comparable to one transition of the multistate model (non-yield to full yield). Through the Rare Event Logistic Regression (RELR) model, this study finds that near crash events are highly associated with a single cross stage, a high number of lanes, and night time. In addition, this study reveals that there is a strong association between partial-yield and near-miss events. Additionally, it is found that for every second that traffic continues to flow while pedestrians are waiting to cross, the probability of a partial-yield event occurring increases by 2.1%, while that of near-crash events increase by about 3%. Moreover, the influence of the crosswalk features and the distance at which drivers yield with respect to the yield line (spatial yielding) was assessed. The logistic regression results for associating drivers’ spatial yielding results shows that the odds for drivers’ spatial yielding are high if the crosswalks are equipped with Rectangular Rapid Flashing Beacons (RRFBs) at the advanced pedestrians crossing signs (APCSs), in the presence of “State Law” and “PED XING” signs. On the other hand, long distances from stripes to the yield lines, multiple cross stages, and high Annual Average Daily Traffic (AADT) are associated with decreased spatial yielding compliance.
Regarding pedestrian-infrastructure interactions, the logistic regression results reveal that the arrival sequence to a crosswalk has the highest impact on warning light activation tendencies. This means that the first arriving pedestrians are eight times more likely to press pushbuttons. Moreover, males, the elderly, children, and teens are less likely to press pushbuttons. Furthermore, pedestrians who are involved in secondary activities, such as carrying/holding objects in their hands, have a relatively low odds ratio of pressing the pushbutton, while phone use is a statistically insignificant factor. Several infrastructure and traffic factors, including flash-based signal types (CRFBs, CFBs and RRFBs), a high number of lanes, residential land use, and higher oncoming vehicle speeds are associated with an increase of pushbutton pressing. Among the models applied for spatiotemporal crossing compliance, the logistic regression outperformed the multinomial logit and the ordered logit models. The logistic regression results reveal that the active WALK signal and a crossing incident involving female(s) only are the factors positively associated with pedestrians’ spatiotemporal crossing compliance. On the other hand, wait time, children, and teens, as well as people who cross while using a phone or riding a bike are negatively associated with spatiotemporal crossing compliance.
Based on the study’s findings, several recommendations are provided. The findings and recommendations from this study are expected to have academic, industry, and community benefits. Planners and engineers can benefit from this study by learning which countermeasures improve safety for both pedestrians and drivers. The models can be used by academicians and other practitioners to assess the scenarios in question. Improved pedestrian safety due to the selection of appropriate countermeasures, which fit a particular location, is a benefit that directly impacts the community.
Crossing compliance; Driver-pedestrian-infrastructure interactions; Pedestrian safety; Signalized midblock crosswalks; Traffic safety; Yielding compliance
Civil Engineering | Transportation
University of Nevada, Las Vegas
Kutela, Boniphace, "Modeling Driver-Pedestrian-Infrastructure Interactions at Signalized Midblock Crosswalks" (2019). UNLV Theses, Dissertations, Professional Papers, and Capstones. 3635.
IN COPYRIGHT. For more information about this rights statement, please visit http://rightsstatements.org/vocab/InC/1.0/