Modelling Crowd Dynamics: a Multiscale, Measure-theoretical Approach

TL;DR

This paper introduces a multiscale, measure-theoretical model of crowd dynamics that unifies macro and micro perspectives, incorporates thermodynamic principles, and demonstrates solution existence, uniqueness, and numerical approximation.

Contribution

It develops a novel measure-theoretical framework coupling micro and macro models, ensuring thermodynamic consistency and providing a robust numerical scheme for crowd behavior simulation.

Findings

Model aligns with thermodynamics via entropy inequality.

Proves existence and uniqueness of solutions for the model.

Demonstrates two-scale micro-macro behavior through simulations.

Abstract

We present a strategy capable of describing basic features of the dynamics of crowds. The behaviour of the crowd is considered from a twofold perspective. We examine both the large scale behaviour of the crowd, and phenomena happening at the individual pedestrian's level. We unify micro and macro in a single model, by working with general mass measures and their transport. We improve existing modelling by coupling a measure-theoretical framework with basic ideas of mixture theory formulated in terms of measures. This strategy allows us to define several constituents of the crowd, each having its own partial velocity. We can thus examine the interaction between subpopulations that have distinct characteristics. We give special features to those pedestrians that are represented by the microscopic (discrete) part. In real life they would play the role of leaders, predators etc. Since we are interested in the global behaviour of the rest of the crowd, we model this part as a continuum. By identifying a suitable concept of entropy, we derive an entropy inequality and show that our model agrees with a Clausius-Duhem-like inequality. From this inequality natural restrictions on the proposed velocity fields follow; obeying these restrictions makes our model compatible with thermodynamics. We prove existence and uniqueness of a solution to a time-discrete transport problem for general mass measures. Moreover, we show properties like positivity of the solution and conservation of mass. Our results are valid for mass measures in their most general appearance. We give a robust scheme to approximate the solution and illustrate numerically two-scale micro-macro behaviour. We experiment with a number of scenarios, in order to capture the emergent qualitative behaviour. Finally, we formulate open problems and basic research questions, inspired by our modelling, analysis and simulation results.