Human behavior modeling for simulating evacuation of buildings during emergencies

Abstract

Every year, a considerably large number of disasters occur as a result of natural events or human faults. In order to decrease the damage and casualties associated with each disaster, it is crucial to get prepared for these kind of situations. Indeed, emergency evacuation is a crucial part of this preparation. Researchers have been working on creating evacuation simulation systems over the past few decades. They are trying to model the environment, human physiology and psychology as realistic as possible to make the analysis more accurate. However, there is yet no comprehensive system which well cover the emerging situations and guarantees the avoidance of causalities, environmental and economic damage. In this paper, we propose an approach which combines a multi-agent model with fuzzy logic to smoothly and successfully handle multiple features of each individual to simulate common human and group behavior during safety egress. The developed simulation system considers situations where a crowd is blocked inside a building or a zone during a disaster. Agents capture various aspects related to humans who may be present in such a region. Each agent possesses different features to realistically simulate a human by encapsulating the psychology, sociology, mood, reaction, etc. Integrating fuzziness in the model allows for more natural capturing of human behavior during the evacuation process. Different scenarios have been tried in the conducted experiments. The outcome revealed the importance of various characteristics of the zone to be evacuated and how they could positively affect the safeness of the evacuation plans. For instance, increasing the width of an exit up to a certain limit may be very beneficial in the process based on the density of the crowd to be evacuated. Actually, the reported simulation results demonstrate the applicability and effectiveness of the proposed approach.