Visualization of Crowd Flow in Large-scale Facility Using Agent-based Simulation

ABSTRACT: To prevent crowd accidents in large-scale facilities that are available to the general public, it is important to share information on safety measures among organizers, the police, the fire department, and government officials. In this study, we present a crowd simulation system that is capable of sharing visual information on the estimated flow of the crowd and that can be used when the safety plan for a facility is being developed. The simulation is agent based and aims to reproduce the macroscopic phenomena of a crowd flow, especially in a queue, by modeling the microscopic interactions between the agents. This crowd behavior model is divided into three parts: route choice, crowd movement, and queue formation. Route choice is implemented using a simple graphical user interface (GUI); this enables the choice of an arbitrary route to guide each agent. Crowd movement is based on a social force model. In addition, we implement a queue formation model in which the route to a destination is changed according to the length of the queue in a specific area. The results are as follows. We calibrated the parameter of the simulation and reproduced the density of a crowd flow based on empirical observations. Queuing behavior was also verified for the emergence of the macroscopic phenomena of stop-and-go waves. Moreover, we applied this simulation to evaluate crowd safety for a special event held in a large-scale commercial facility. We propose a new method for safety experts and non-experts to share visual information about the flow of crowds.