Modeling crisscross motion dynamics of high-density crowd: Coupled directional selection and shoulder avoidance behaviors

Crowd motion is a self-organizing phenomenon driven by the interplay between individual self-propulsion and interpersonal interactions, with different relative strengths to generate distinct dynamic crowd structures. The classical Social Force Model (SFM) assumes pedestrians are homogeneous, self-driven particles. A position-adapted collision avoidance mechanism is used to predict sparse crowd movement patterns. However, SFM fails to account for the heterogeneity in human body structure and its impact on interpersonal interactions, while the latter dominates motion dynamics of high-density crowds. On the basis of the tri-circle model, this study introduces an improved shoulder mechanism into the counter-flow model. Using this enhanced SFM, the crisscross motion dynamics of the high-density crowd is solved numerically by considering three kinds of intersection angles of initial crowd movement directions, that is, diagonal path of cross-flow with θc = 90°, parallel path of counter-flows with θc= 180° and the mixed crisscross scenarios with θc = 90° ⋀ 180°. The simulation results reveal that pedestrians can utilize the new freedom of body rotation in congested situations, enabling it to fully utilize the flexibility of body rotation, thereby establishing a more coordinated spatial structure in high-density populations and simulating crowd movement posture more realistically. The developed model provides a valuable understanding of crowd dynamics and is beneficial for optimizing evacuation strategies.