Spatial fluctuation on speed-density relationship of pedestrian dynamics

Abstract

A position-dependent estimation approach for pedestrian fundamental diagrams (speed-density relationship) is proposed. Probabilistic approaches have succeeded in explaining fluctuations contained in the empirical pedestrian flow data by handling differences among pedestrians. Dealing with differences between positions is also important in the evaluation of pedestrian facilities. The fluctuation over positions is modelled as a spatial dependence structure. The eigenvector spatial filtering approach combines this structure with a non-spatial fundamental relationship. We show the superiority of the proposed approach over a non-spatial model through two applications of real pedestrian stream data. Furthermore, the estimated result shows a spatial fluctuation pattern over the experimental area.