Potential field cellular automata model for overcrowded pedestrian flow

Abstract

A cellular automata model for overcrowded pedestrian flow is proposed by dividing a normal cell into nine small cells, among which the compressibility is such that one pedestrian referred by a central cell can share a side or corner cell with at most one or three other pedestrians. The compressibility between a pedestrian and a wall is half as much as that between pedestrians. Hence the model allows a density exceeding 10 ped/m2. The direction of motion minimizes the deviation from the negative gradient of cost potential, and the probability of movement decreases with the deviation and travel cost. The simulated fundamental diagram and evacuation process generally agree with field studies in the literature. The increased cost helps reproduce the formation of lanes in counter-flow; however, the degree depends on the amplitude of increase and the density, which is measured by the probability of gridlock and order parameter through simulation.