📝 SummarXiv

Abstract

This paper develops a general force-based pedestrian model named CosForce, in which cosine functions are employed to describe asymmetric interactions. These functions implicitly capture the mechanisms of anticipation and reaction. By focusing on binary interactions, the CosForce model provides new insights into pedestrian modeling while achieving property of linear time complexity. Two specific scenarios in crowd dynamics were analyzed: self-organization (entropy decrease) and crowd collapse (entropy increase). The average normalized speed and order parameter were introduced to quantitatively describe the processes of crowd dynamics. Quantitative evaluations demonstrate that phase separation in crowds is effectively reproduced by the model, including lane formation, stripe formation, and cross-channel formation. Next, in the simulation of mass gathering, within a density-accumulating scenario, processes of critical transition in high-density crowds are clearly revealed through time series observations of the order parameter. These findings provide valuable insights into crowd dynamics.