Modeling and prediction of driver-vehicle-unit velocity using adaptive neuro-fuzzy inference system in real traffic flow / Iman Tahbaz-zadeh Moghaddam...[et al.]

Prediction of the driver-vehicle-unit (DVU) future state is a challenging problem due to many dynamic factors influencing driver capability, performance and behavior. In this study, a soft computing method is proposed to predict the accelerating behavior of driver-vehicle-unit in the genuine traffic stream that is collected on the California urban roads by US Federal Highway Administration’s NGSIM. This method is used to predict DVU velocity for different time-steps ahead using adaptive neuro-fuzzy inference system (ANFIS) predicator. To evaluate the performance of proposed method, standard time series forecasting approach called autoregressive (AR) model is considered as a rival method. The predictions accuracy of two methods are compared using root mean square error (RMSE), mean absolute percentage error (MAPE) and coefficient of determination or R-squared (R2) as three error criteria. The results demonstrate the adequacy of proposed algorithm on real traffic information and the predicted speed profile shows that ANFIS is able to predict the dynamic traffic changes. The proposed model can be employed in intelligent transportation systems (ITS), collision prevention systems (CPS) and etc.