Exploring the Dynamic Relationship between Pushing Behavior and Crowd Dynamics

Abstract

Crowds, subjects of considerable complexity, have been extensively studied both as homogeneous entities and as collective sums of individual movements in various studies. However, crowd models, being grounded in physics, are limited in terms of incorporating psychological perspectives on individual behavior. Building upon the premise that crowd behavior is heterogeneous and dynamic, particularly in bottleneck scenarios, this study aims to explore the nuances of forward motion. Adopting the category system proposed by Lügering et al. (2022) (consisting the following categories: strong pushing, mild pushing, just walking, falling behind), this paper investigates the circumstances and locations where pushing or non-pushing behaviors arise, intensify, or cease within crowds approaching bottlenecks. The study utilized 14 video materials obtained from previous laboratory pedestrian experiments to examine the spatial characteristics of forward motion and pushing behavior in relation to corridor widths and varied motivational instructions. Two trained raters independently annotated these videos, achieving satisfactory inter-rater agreement (KALPHA = .65) , and a joint dataset was then created for each video. These videos consisted both high (7 videos) and low (7 videos) motivation scenarios. The importance of corridor width was also considered: four videos featured a 5.6m width, another four featured a 4.5m width, and the remaining videos displayed widths of 3.4m, 2.3m, and 1.2m twice. Our findings suggest a tendency for increased pushing behavior or an increase in the categories as individuals approach the bottleneck, regardless of the width of the corridor or the motivational instruction. Furthermore, non-pushing behaviors were predominantly observed in the areas farther away from the bottleneck. A noticeable trend was observed in high motivation scenarios, which generally exhibited more instances of pushing behavior. The effect of corridor width indicated that, in certain cases, pedestrians who push in wider corridors experience faster access to the bottleneck. However, this effect is less significant in narrower widths.