Modelling the driving speed on expressway ramps based on floating car data

Abstract

Driving speed is one of the key concepts and risk factors in transportation research. The insights into operational and safety aspects of driving can be provided by floating car data (FCD), collecting information about speed, position and time by vehicles themselves. FCD are often recorded at high frequency, representing a continuous phenomenon. As such, convenient approach can be functional data analysis (FDA). Speed trajectories are investigated, identifying sections with rapid changes of speed and sections where drives in central and auxiliary lanes cannot be distinguished in terms of their speed. The effect of curvature and auxiliary lanes on driving speed was shown by functional regression models. Significant influence of road shape on speed was found for ramps with complex shape. The accuracy of regression models was assessed by RMSE, NRMSE, and precision of estimators by point-wise confidence intervals. The analysis was performed on an expressway interchange in Brno, Czech Republic.