A Bayesian inference approach to the development of a multidirectional pedestrian stream model

Abstract

In this paper, we develop a mathematical model to represent the conflicting effects of multidirectional pedestrian flows in a large crowd. The model is formulated based on Drake's model of traffic flow. Rather than relate the speed of a pedestrian stream solely to the pedestrian density, we introduce the flow ratio and intersecting angle between streams as variables. To calibrate the model, data collection was conducted through the video recording of pedestrian movements on a pedestrian street in Mong Kok, Hong Kong. Bayesian inference was adopted to calibrate the parameters based on the information from a previous experiment. Finally, we study the relationships among the speed, density, flow and intersecting angles of the pedestrian streams and predict how these variables affect the pedestrian movements.