Advisory Committee Chair
Virginia P Sisiopiku
Advisory Committee Members
Michael Anderson
Akhlaque Haque
Jason Kirby
Robert W Peters
Document Type
Dissertation
Date of Award
2015
Degree Name by School
Doctor of Philosophy (PhD) School of Engineering
Abstract
There has been a growing emphasis on improving multimodal serviceability of transportation systems. The integration of pedestrian behavior in mainstream traffic models, hence, is gaining traction. Advances in modeling pedestrian-vehicle interactions can aid in better understanding and development of control strategies for pedestrian safety. In support of such efforts, this study delivers an improved understanding of pedestrian and driver behavior at crossing locations through the development of discrete gap choice models to model the interaction between pedestrian and drivers at unsignalized crossing locations. First a comprehensive literature review was performed and latest advances in modeling pedestrian-vehicle interaction at various types of crossing locations were evaluated and summarized. Lessons learned from earlier studies and needs for further research were identified and presented. Next, traffic data were collected at 27 midblock crossings sites located in Alabama, Florida, and North Carolina using video cameras and laser speed radars. Data reduction and analysis followed and produced macroscopic performance measures such as the percentage of driver yielding, average observed speeds, pedestrian delay, and observation of pedestrian-vehicle conflicts. The data were then used for development and validation of discrete choice models for pedestrian-vehicle interaction at midblock crosswalks. The research outcome is a robust, standalone model of pedestrian gap acceptance behavior at midblock crossings sensitive to the available gap length and distinguishing gap and lag events. Additional data collection was performed at Two Way Stop Controlled (TWSC) intersections in Birmingham, AL and the data were used to develop discrete choice models using probit and logit based formulations for TWSC marked crosswalks. In a first of its kind study, impacts of visibility due to time-of-day conditions were assessed for TWSC locations. The results show that pedestrians waiting to cross the street were more likely to accept an available gap in the daylight conditions compared to under low visibility conditions during dark evening hours. The research reported in this paper offers new, robust pedestrian gap acceptance models for midblock and TWSC pedestrian crosswalks that can be used in traffic operational analysis. These can be further incorporated in micro-simulation packages thus improving the accuracy of pedestrian behavior modeling at unsignalized crossings in the future. Better understanding of pedestrian-vehicle interactions at unsignalized crosswalks is expected to lead to improved treatments that would, in turn, enhance traffic safety at pedestrian crossing locations.
Recommended Citation
Mamidipalli, Shrikanth, "Robust Discrete Choice Models For Pedestrian-Vehicle Interactions At Unsignalized Locations" (2015). All ETDs from UAB. 2376.
https://digitalcommons.library.uab.edu/etd-collection/2376