# An Analysis of Pedestrians' Behavior in Emergency Evacuation Using Cellular Automata Simulation

## TL;DR

This paper uses Cellular Automata simulation to analyze how pedestrians' behavior, including herd effect and environmental awareness, impacts emergency evacuation efficiency and safety.

## Contribution

It introduces a CA-based model that considers dynamic environmental changes and psychological factors, providing new insights into evacuation strategies.

## Key findings

- Herd effect can hinder evacuation efficiency.
- Timely information about environmental changes improves evacuation outcomes.
- Dynamic environment modeling is crucial for realistic evacuation analysis.

## Abstract

To minimize property loss and death count in terror attacks and other emergent scenarios, attention given to timely and effective evacuation cannot be enough. Due to limited evacuation resource, i.e., number of available exits, there exists interdependence among pedestrians such as cooperation, competition and herd effect. Thus human factors - more specifically, pedestrians' behavior in emergency evacuation - play a significant role in evacuation research. Effective evacuation can only be reached when route planning are considered in conjunction with psychological dynamics, which is often ignored. As another drawback, previous research assumes the environment including available exits as stationary. However, we note that during emergency, some exits which are not often utilized in normal times are opened, which potentially helps if pedestrians are aware of them. In this paper, we analyze the effect of pedestrians' behavior, i.e., herd effect and knowledge of changing environment with Cellular Automata (CA) simulation. Results of the simulation show the harmful effect of herd effect as well as highlight the importance of timely informing pedestrians of environmental change. Accordingly, we propose policy and procedural recommendations for emergency management of large, crowded structures. Our future work includes considering more human factors and applying our model to log data provided by videos in public venues, which can further show effectiveness of our model in real scenarios.