Analysis of the homogeneity of driver behaviors at intersections

Abstract

Intersections are road traffic accident-prone locations. About 10–40% of the traffic accidents take place at intersections worldwide every year. An intersection in Haidian District, Beijing was surveyed by video survey method, and the survey videos were processed and analyzed. The velocities and trajectories of the vehicles at the intersection were collected. Analysis of the homogeneity of driver behaviors shows that vehicles have the best homogeneity of velocities when passing the stop line and the crosswalk. Right-turning vehicles have the best homogeneity of velocities, left-turning vehicles take the second place, and homogeneity of velocities of straight-going vehicles is the worst. Straight-going vehicles have the best homogeneity of trajectories, right-turning vehicles take the second place, and left-turning vehicles have the worst homogeneity of trajectories. The results provide references for driver behavior modeling at intersections and the development of intersection driving assistant systems.