Denis Ullmo (LPTMS)

Pedestrians in static crowds are not grains, but game players

The short-term (‘operational’) dynamics of pedestrian crowds are generally thought to involve no anticipation, except perhaps the avoidance of the most imminent
collisions. I will show that current models rooted in this belief fail to reproduce essential features observed experimentally by Nicolas et al. [Sci. Rep. 9, 105 (2019).] when a static crowd is crossed by an intruder.

The missing ingredient can be identified as the pedestrians’ ability to plan ahead far enough beyond the next interaction. On this basis, I will introduce a minimal model based on mean-field game theory which proves remarkably successful in capturing the experimental observations associated with this setting, but also other daily-life situations such as partial metro boarding. These findings are clear evidence that a long term game theoretical approach is key to capturing essential elements of the dynamics of crowds.

[refs : Phys. Rev. E 107, 024612 (2023), SciPost Phys. 16, 104 (2024)]