Model of encroachment into opposite lanes in horizontal curves of rural roads

Abstract

Horizontal curves of rural roads are accident-prone segments of the route. Sharp curves, steep slopes, and reduced visibility due to the mountainous environment greatly affect the driver behavior and performance. Lane-keeping ability, which is quite crucial in head-on road collisions, is a lateral driver behavior examined in a number of previous studies. This study, which is aimed to examine the naturalistic behavior, has employed the “aerial video recording” to investigate the drivers' lane-keeping ability in horizontal curves.

To address the risk of encroachment (enc) into the opposite lane, this paper has developed a logistic regression model to predict the probability of a head-on collision with an enc > 0 cm threshold by exploring the relationships between road features (geometric, traffic, pavement conditions, etc.) and driver encroachment into the opposite lane. To this end, use was made of the data of 785 vehicles in 11 horizontal curves (in Kashmar-Neyshabor and Siahkal-Deylam mountainous routes) with radii in the 30–150 (m) range, deflection angles in the 80°-150° range, and slopes in the 0–8% range. The explanatory variables used in the model included the start point position (sp), road slope (Gr), sufficient stopping sight distance (Sd) and difference between the posted and vehicle speeds in mid-curves (DPS). According to the results, speeding and curve rising of 70° increased the encroachment probability, and steep upgrades exacerbated it; at a sufficient stopping sight distance, it reached 85%.