Towards realistic and effective Agent-based models of crowd dynamics

摘要

The authors propose a new methodology for creating realistic and effective models of crowd dynamics, which takes into account the Agent-based approach combined with non-homogeneous and asynchronous Cellular Automata. The proposed methodology makes it possible to model pedestrians' dynamics in complex environments, like stadiums or shopping centers, and enables mimicking of the pedestrians' complex decision making process on different levels: strategic and tactical/operational. On the other hand, the use of the Agent-based approach makes it possible to apply different scenarios and situational contexts, namely competitive and non-competitive evacuation or free movement of pedestrians. The proposed approach was tested in large-scale test cases, namely the evacuation of the Allianz Arena football stadium in Munich and other stadiums like Wisla Krakow Stadium or GKS Tychy Stadium, as well as AGH University facilities.