Modeling car-following behavior as instantaneous maximization of drivers' utility for estimation of traffic level of service

This study points out the importance of estimating the traffic service level. For the estimation, a car-following model is proposed based on the assumption that a driver tries to adjust acceleration to maximize utility instantaneously. Two types of data are applied for model verification: data from a probe car that are precise but probably biased, and data from a video camera which are not expected to be very precise. The fundamental performance of the car-following model is verified with data from the probe car. The model generally fits to drivers' data taken from video image, and individual differences - which are probably not negligible when estimating drivers' satisfaction - in the parameters of the model is revealed. An experimental estimation of chronological changes in drivers' utility caught by video footage is tested as an application of this model.