Modeling of pedestrian gap acceptance for improving safety at uncontrolled mid-block crosswalks

Pedestrian accidents at uncontrolled mid-block crossings pose a serious risk in the U.S. and other countries. A clear understanding of the parameters causing these accidents is needed to make the crosswalks safer. The purpose of this paper is to develop realistic models for pedestrian gap acceptance behavior at uncontrolled two-lane mid-block crosswalks. Different methodologies for modeling Pedestrian Gap Acceptance (PGA) are proposed. Three different methodologies are employed for studying PGA, the first model is deterministic and solely depends on the gap sizes that are accepted or rejected by pedestrians. The second model is probabilistic and the probability of accepting a gap is calculated as a random variable from a distribution that best fits the data. The third model uses a binary logit approach; multi-attribute regression analyses are performed to capture the decision making process of the pedestrian. Field studies were conducted to collect the data of different attributes using a set of video cameras at several typical unsignalized mid-block crosswalks during two pedestrian flow peak hours of five working days. The data was used to calibrate and validate the different proposed models. The results show that PGA binary logit models perform better than the other models. Moreover, additional insights are provided into the PGA behavior based on the data.