Development of pedestrian crossing behavior and safety index models at signalized intersections under mixed traffic conditions

Pedestrians are considered the most vulnerable road users at intersections, and their crossing behaviors significantly impact the perception of safety. But it is very difficult to analyze pedestrian crossing behavior due to pedestrians' complex and diverse nature. Therefore, this study aimed to develop pedestrian crossing behavior and safety index models by combining qualitative and quantitative data at signalized intersections, which will help improve pedestrian movement. The data were collected and extracted from a videographic and user perception survey at four signalized intersections in Coimbatore, India. The variations in pedestrian crossing behavior concerning socio-economic variables, travel attributes, pedestrian and traffic characteristics, and geometric characteristics were identified using Pearson and Spearman correlation tests. The binary logit models were developed to evaluate pedestrian crossing behavior in noncompliance with signal, crossing in an unmarked area, and mobile usage while crossing the crosswalk. A structural equation modeling approach was adopted to combine three behaviors for analyzing crossing behavior. An ordered probit model was developed to estimate the pedestrian perception-based safety index value at signalized intersections. Finally, all the models were validated, and the statistical results showed that the proposed models predicted pedestrian crossing behavior and the safety index value more accurately under mixed traffic conditions.