An iteration method for predicting vehicular trajectory based on car-following models

In the recent two decades, the improvement of vehicular network gave momentum to the research of vehicle collision warning systems, which focus on curbing wounded and fatalities in crashes, increase the requirement to uncover the mechanism of driving. Car-following algorithm, a series of methodology using formula to express the interaction between drivers and vehicles, have reproduce several intricate phenomenon. In this paper, some highly-cited car-following algorithm are calibrated based on the positional information in real traffic and exploited to predict future trend of vehicles more precisely. By analyzing the error distribution in prediction vehicles' trajectory for each calibrated car-following models, an iteration method is used to mitigation positional errors and thus may help to resolve the errors caused by anticipation or stochastic characteristic of drivers. The results indicate optimistic performance of proposed trajectory prediction model especially in constant car-following situation.