Evaluation of Traffic Parameters from Real Trajectories by Use of PCA

Abstract

This paper shows how principal components analysis (PCA) can be used to analyze a large set of real trajectories of vehicles on a bend. PCA gives an opportunity to reduce the mathematical and numerical operations necessary to extract real values of traffic parameters (such as curvature, centrifugal acceleration, and the rate of change of this acceleration) from the set of trajectories. Also profiles of these parameters can be obtained. Applied to an exemplary set of real trajectories, PCA enabled finding that maximum curvature and rate of change of centrifugal acceleration are square functions of vehicle velocity, and maximum acceleration is its linear function.