Revealing the hidden rules of bidirectional pedestrian flow based on an improved floor field cellular automata model

Abstract The floor field cellular automaton model (FFCA) has been widely adopted to simulate pedestrian and evacuation dynamics. Many self-organized phenomena could be reproduced with the FFCA model, such as the lane formation in bidirectional pedestrian flow. However, as presented in this study, when we tried to use the FFCA model to simulate an experiment on bidirectional pedestrian flows performed in discrete space and time, we found the model failed to agree with the empirical results. The gridlock formation (not observed in the experiment) was unavoidable in the FFCA model and the clearance time in simulation was much larger than that of the experiment. From the experiments, we observed that people would like to stop if they would foresee a benefit in the near feature and consequently give way to people coming from the opposite direction. This inspired us to incorporate such behavioral rules for modeling pedestrian movements in bidirectional flows. To this end, we introduced a waiting time rule to the original FFCA model. Results showed the performance of the model could be significantly improved. The gridlock probability could be reduced to zero, with clearance time agreeing well with the experimental outcome. Findings from this study can provide meaningful insights for researchers into understanding the pedestrian behavior in bidirectional flow and help develop more reliable simulation software.