[Paper Review] A macroscopic crowd motion model of gradient flow type

Bertrand Maury, Aude Roudneff-Chupin+1 more|arXiv (Cornell University)|Feb 3, 2010

Geometric Analysis and Curvature Flows28 references4 citations

TL;DR

This paper introduces a macroscopic crowd motion model where individuals' desired velocities are projected onto feasible velocity sets to enforce incompressibility and congestion constraints. Using a gradient flow structure in the Wasserstein space, it establishes well-posedness despite the functional's lack of geodesic convexity and finite values, offering a new framework for modeling dense pedestrian flow with strong non-overlapping constraints.

ABSTRACT

A simple model to handle the flow of people in emergency evacuation situations is considered: at every point x, the velocity U(x) that individuals at x would like to realize is given. Yet, the incompressibility constraint prevents this velocity field to be realized and the actual velocity is the projection of the desired one onto the set of admissible velocities. Instead of looking at a microscopic setting (where individuals are represented by rigid discs), here the macroscopic approach is investigated, where the unknwon is the evolution of the density . If a gradient structure is given, say U is the opposite of the gradient of D where D is, for instance, the distance to the exit door, the problem is presented as a Gradient Flow in the Wasserstein space of probability measures. The functional which gives the Gradient Flow is neither finitely valued (since it takes into account the constraints on the density), nor geodesically convex, which requires for an ad-hoc study of the convergence of a discrete scheme.

Motivation & Objective

• To develop a macroscopic model of crowd motion that enforces strong non-overlapping constraints, avoiding the relaxed treatment common in most existing models.
• To formulate the crowd motion problem as a gradient flow in the space of probability measures equipped with the Wasserstein distance.
• To handle the non-convex and non-finite-valued nature of the underlying functional through an ad-hoc analysis of discrete schemes.
• To establish a connection between microscopic contact dynamics and macroscopic incompressible flow by projecting desired velocities onto feasible sets.
• To explore the model's applicability beyond pedestrian dynamics, particularly in cell dynamics and granular media, where similar constraints arise.

Proposed method

• Model the crowd as a density function ρ(t,x) evolving via the continuity equation ∂tρ + ∇·(ρu) = 0.
• Define the desired velocity field U(x) = −∇D, where D is the distance to the exit, representing individual preferences.
• Enforce the incompressibility and non-overlapping constraints by projecting the desired velocity U onto the set of admissible velocities Cρ, yielding the actual velocity u = P_{Cρ}(U).
• Formulate the problem as a gradient flow in the Wasserstein space of probability measures, with a functional that incorporates both the transport cost and the constraints.
• Use a discrete time scheme to approximate the gradient flow, proving convergence via a tailored analysis due to the functional's lack of geodesic convexity.
• Apply optimal transport theory, particularly the Wasserstein distance, to define the geometry of the space in which the flow evolves.

Experimental results

Research questions

• RQ1 How can a macroscopic crowd motion model enforce strong non-overlapping constraints without relying on relaxed repulsive forces?
• RQ2 Can the crowd motion problem be formulated as a gradient flow in the Wasserstein space when the underlying functional is neither finite-valued nor geodesically convex?
• RQ3 What is the role of the velocity projection operator P_{Cρ} in ensuring incompressibility and congestion avoidance in the macroscopic limit?
• RQ4 How does the macroscopic model compare to microscopic models in capturing phenomena like arching or jamming near exits?
• RQ5 In what contexts beyond pedestrian evacuation—such as cell dynamics or granular flows—can this framework be meaningfully applied?

Key findings

• The model successfully formulates crowd motion as a gradient flow in the Wasserstein space, even though the underlying functional is not geodesically convex and takes infinite values on certain densities.
• The discrete time scheme used to approximate the gradient flow is proven to converge, despite the lack of standard convexity assumptions, through a specialized analytical framework.
• The macroscopic model does not reproduce microscopic phenomena such as arching near exits, due to its reliance on local density rather than local contact structure.
• The model can be extended to other systems, such as a unilateral version of the Keller-Segel equation, where cells move toward a chemoattractant but are constrained by a maximum density.
• The model provides a first-order macroscopic counterpart to second-order pressureless gas models with congestion, suggesting a path toward higher-dimensional granular flow modeling.
• The characteristic function of a single ball is a static solution in R³, indicating that mass concentration is naturally prevented by the congestion constraint.