Simulating the bi-directional pedestrian flow under high densities by a floor field cellular automaton model

In this paper we propose one floor field cellular automaton model, which can simulate the bi-directional pedestrian flow at high densities. Based on the model rules proposed by Nowak and Schadschneider, we make some modifications, including changes of cell size, realistic velocity configurations and extended lateral movement. The best parameters are determined by the results of sensitivity analysis. Some data extracted from pedestrian flow experiments, including the fundamental diagrams and the lane numbers after lane formation, are used for model validations. When the corridor width is 2m, the simulation results show that the fundamental diagrams of bi-directional flow are quantitatively similar to the experimental results, and the lane formation could be successful at high densities. By comparing with the statistical results in the experiments, we find the probabilities of the occurrence of different lanes are similar to the experimental results: two-lane and three-lane states usually emerge, while the probability for more lanes is low. The inhomogeneous distributions of pedestrians also could be observed. We think our model could be a good choice for simulating the lane formation process at high densities.