Agent-based Modeling of Pedestrian Behavior at an Unmarked Midblock Crossing

:Illegal pedestrian behavior is considered one of the main reasons for pedestrian-related crashes. This paper provides a detailed outlook of the gap acceptance behavior of pedestrians illegally crossing at an urban midblock section of a six-lane road. A simulation model was developed to capture the illegal crossing behavior for the pedestrians using agent-based modeling techniques and implemented in Anylogic. Once calibrated to study the crossing behavior, several distributions were fed into the model to determine the best distribution that mimics the pedestrians' movements. It was found that the normal distribution is the closest fit. The results suggest that the pedestrians add a factor ranging from 1.25-1.5 to vehicle speeds before anticipating the gap, then determine whether to cross or not. The simulation tool was found useful for replicating the illegal crossing behavior. The results from this study can be helpful in assessing the pedestrian safety and providing better crossing facilities for pedestrians in urban areas.