Performance analysis of urban drivers encountering pedestrian

According to the World Health Organization (WHO), Iran is one of the countries with the highest rates of road deaths in the world. About 18,000 people die every year in road accidents in Iran, where about 22 percent of the dead are pedestrians. The purpose of the present paper is to investigate the effective factors on performance of drivers during interaction with pedestrians. Therefore, the Naturalistic Driving Study (NDS) of behaviour of 66 participants (29 males and 27 females, 18-65 years old) has been evaluated in Babol city, Mazandaran province, during 2014-2016. The behavioral studies of the participants were conducted in 216 cases of vehicle-pedestrian interaction in divided road and 485 cases in undivided road through video-recorded process. The results showed that vehicle speed and distance to pedestrians are the most important factors affecting the occurrence of vehicle-pedestrian interaction in both sites. Moreover, the results show that on the divided road, Running when crossing the street by pedestrians, as well as listening to the music by drivers, increases the possibility of interaction. Also, on the undivided road, the attention to the traffic flow of road before crossing by pedestrians, as well as crossing the street in a group increase the probability of driver's performance. Drivers responded to pedestrian crossing the street by some performances such as decreasing speed, changing the line, stop and acceleration collision. Finally, the probabilistic models of driver performance as well as the type of performances based on the variables affecting the behavior of drivers are determined using the binary logit model and multinomial logit model, respectively. Further model validation and transferability were checked and it has been observed that the driver performance and types of performance models developed in this study represents quite well. The inference of these models will be useful to assess Drivers' behavioral models and suggest automotive assistance systems for improving pedestrian safety. (C) 2019 Elsevier Ltd. All rights reserved.