A stochastic optimal velocity model for pedestrian flow

Abstract

We propose a microscopic stochastic model to describe 1D pedestrian trajectories obtained in laboratory experiments. The model is based on optimal velocity (OV) functions and an additive noise determined by the inertial Ornstein-Uhlenbeck process. After statistical estimation of the OV function and noise parameters, we explore the model by simulation. The results show that the stochastic approach gives a good description of the characteristic relation between speed and spacing (fundamental diagram) and its variability. Moreover, it can reproduce the observed stop-and-go waves, bimodal speed distributions, and nonzero speed or spacing autocorrelations.