Car-following models has been studied for decades, and many kinds of models such as desired measures model, safety distance model have been developed maturely. In recent years, the artificial potential field theory has been widely used in path planning. For its good performance in collision avoidance, this paper presents a car-following model based on the artificial potential field theory, aim to describe the movement of the vehicle precisely. Firstly, an artificial potential field function is established, which is composed by velocity potential, side potential, lane potential and obstacle potential. And the acceleration of the vehicle is derived. Then, the calibration and validation are conducted using NGSIM data, with the artificial bee colony algorithm to calibrate the parameter of the proposed model. The root mean square normalised error (RMSNE) is selected to evaluate the results. The assessment shows the proposed model have good performance in describing the car-following behavior of the vehicle. As a result, not only does the artificial potential field theory can be used in machine controlling, it can also be used to describe the movement of real traffic.

CICTP