EMPIRICAL ANALYSIS OF DRIVERS' CAR-FOLLOWING HETEROGENEITY BASED ON VIDEO IMAGE DATA

Microscopic road traffic flow simulation models are increasingly being used to express the dynamic nature of traffic flow. These microscopic traffic simulations require vehicle movements to be modelled and calibrated. Since individual vehicles exhibit heterogeneity, the parameters within these submodels must incorporate some disturbances. However, due to limited observational data, the various parameters among individuals have not been well validated. We collected video data from a 600m section in a merging area of an urban expressway over 1 week using 11 video cameras. By tracking individual vehicles on-screen, we were able to obtain vehicle trajectories. We applied this manually obtained data to investigate individual heterogeneity of vehicle movement; in particular, we focused on car-following behaviour, analysing heterogeneity among individuals. We also examined relationships between car-following behaviour and traffic conditions, driving lanes, and road geometry.