Agent-based continuous-space particle pedestrian model

Abstract

In the design and evacuation evaluation of highly crowded places such as metro stations, continuous-space particle models have gained increasing attention because they are capable of simulating normal crowd self-organisation phenomena including lane formation without the need for complex walking rules. However, such models do not consider pedestrians' high level of automatic control. The paper introduces a new method to the original social force model based on a smart agent model using object-oriented programming. A corresponding implementation method is formulated and a corresponding algorithm designed. The parameters of the pedestrian body, the walking characteristics and the model itself are calibrated, and the modified model is implemented on a VC++ 2008 platform. Both one-directional and bidirectional interchange walkway simulations are studied. The results demonstrate better lane formation and the fundamental diagram from the simulation fits well with that obtained from a field study.