Simulating Pedestrians in Evacuation Processes: A Novel Approach

Abstract

Pedestrian simulation is a central issue in evacuation related topics; an issue that has recently received renewed interest. In order to estimate escape time from a building, this paper describes a two-module model which combines Agent-Based Models (ABM) and Cellular Automata (CA). The former model (ABM) simulates pedestrians exploring the building space; the latter (CA) simulates the proper evacuation process. The novelty of the model is represented by the first module’s approach, which is inspired to Ant Colony Optimization (ACO). Using this metaphor, it is possible to simulate the way in which people draw their cognitive map of the building’s space. According to ACO, agents represent “scout ants” looking for the exit. Initially, ants move in random fashion. When an ant reaches the exit, it updates the grid by adding an amount of pheromone. The result is a pheromone trail that follows the shortest possible path from anthill to the exit cell. Running the former module, a map containing distances from each exit point to the exit is obtained. The latter CA module uses this map to estimate escape time.