Modeling and prediction of driver preparations for making a right turn based on vehicle velocity and traffic conditions while approaching an intersection

This paper focuses on modeling and predicting the influence of a vehicle’s velocity and the relative position between a driver’s vehicle and a vehicle to the front or rear on the onset of driver preparations for making a right turn at an intersection. Repeated experiments were carried out on a public road to measure driver preparations, including releasing the accelerator pedal, moving the right foot to cover the brake pedal, and activating the turn signal, as well as to record vehicle velocity and the relative distances to the leading and following vehicles. Structural equation modeling was applied to estimate these relationships quantitatively. Two separate latent variables accounting for the interaction between the driver’s vehicle and leading or following vehicles, and the onset point where driving behavior changes from straight mode to preparation mode before a right turn, were introduced in the specification of the structural equation model. The model estimates and testing indicate that the proposed structural equation model represents well the relationship hypothesized by observational analysis: that the vehicle velocity and the traffic conditions surrounding the driver’s vehicle strongly influence the driver’s deceleration and more weakly influenced turn signal activation. The proposed structural equation model contributes to the prediction of the onset locations of covering the brake pedal and activating the turn signal based on the vehicle velocity and the relative distances to the front and rear vehicles when the accelerator pedal is released. The prediction accuracy is high compared with predictions in which the vehicle velocity and the headway or rear distances are not taken into account or predictions using a single regression model with one independent variable, namely driving speed. Finally, a possible new addition to in-vehicle navigation systems that detects unusual driver behavior by predicting the driver’s preparatory maneuvers is discussed.