Analysis and Predictive Modeling of the Stochastic Pedestrian Behavior Patterns at Signalized Intersections for Safer Intersection Design and Control

Pedestrianvehicle conflicts are considered as one of the most common safety problems in urban areas. Most of these conflicts are located at intersections where pedestrians and vehicles share the same space. In the Kingdom of Saudi Arabia, although pedestrian-vehicle crashes are about 2% from total recorded crashes, these crashes are classified as the most severe and fatal ones (Kingdom of Saudi Arabia, National Crash Statistics Report, 2009). In Japan, more than one-third of the total traffic crash fatalities are pedestrians at crosswalks (Accident Statistics in Japan, 2011) while it is about 15% in Germany (German Institute for Economic Research, 2010). Although insufficient attention has been given to pedestrians and their facilities including crosswalks in KSA, it is expected that this trend will completely change especially with the huge investment in public transport including metro, rail and bus systems. This requires efficient pedestrian facilities from crosswalks, sidewalks and walkways to provide safe and smooth access to transit stations. It is also important to mention that although walking is not a common mode of transport in KSA, still large pedestrian volumes are observed in the Central Business District CBD areas of large cities as well as near main activity areas such as holly places and shopping centers. When analyzing pedestrian vehicle conflicts, it is concluded that the period after the end of pedestrian green interval PG (the onset of pedestrian flashing green interval PFG) is the most critical since drivers do not expect pedestrians to start crossing meanwhile pedestrians hurry up to cross before the red indication is displayed (Iryo-Asano et al., 2013). Thus, it is essential to understand pedestrian behavior during this interval to develop safe and efficient signal control strategies for intersections. This study aims to analyze and model the probabilistic behavior of individual pedestrians that are approaching crosswalks and may start crossing in KSA and to compare this behavior with that of pedestrians in Japan. Furthermore, an international review of design concepts of pedestrian signal settings will be conducted. The rationality of the existing pedestrian signal settings in KSA will be assessed and pedestrian compliance will be investigated. For that, several signalized crosswalks in the CBD of Dammam-Khobar area will be selected for video recording. Simultaneously video data for signalized crosswalks with similar characteristics in Japan will be collected. TrafficAnalyzer image processing program will be used to extract individual pedestrian maneuvers. Pedestrian behavior is mainly divided into crossing decision (stop-go) and walking speed. Microscopic pedestrian model will be proposed to represent individual pedestrian maneuvers. Using evolutionary computation techniques such as genetic algorithms, the proposed microscopic model will be tuned to best fit individual pedestrian profiles. The distribution of the estimated parameters will be analyzed and modeled using suitable probabilistic models such as gumbel, normal or gamma as functions of various influencing factors including crosswalk geometry, signal timing, conflict with turning vehicles, gender and age (if needed data can be collected). Developed models can be used for the estimation of clearance times, frequency of illegal crossing, and speed distribution. They can be utilized to propose safer pedestrian signal settings. Moreover, the planned international comparison of pedestrian behavior is very useful, since it can highlight the impacts of different pedestrian signal control practices (timings and indications) and cultural customs.