Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A dynamic state-based model of crowds

Martyn Amos, Steve Gwynne, Anne Templeton

TL;DR

Addressing the need to describe time-evolving crowd behavior beyond static typologies, the paper introduces a dynamic state-based model inspired by statecharts, organized into Assembly, Mode, Structure, and Dispersal phases. It encodes crowd evolution as threads of states, enabling sub-crowd emergence and cross-phase transitions without assuming underlying psychology, and it notes extensions such as transition weights and persistent tags for memory. The framework is designed to be descriptive and integrable with existing simulation approaches for both normal and incident crowds, improving the ability to report on real events and to study safety implications. The approach offers a flexible, extensible standard for capturing macroscopic crowd dynamics and their time-varying configurations across diverse environments.

Abstract

We consider the problem of categorizing and describing the dynamic properties and behaviours of crowds over time. Previous work has tended to focus on a relatively static "typology"-based approach, which does not account for the fact that crowds can change, often quite rapidly. Moreover, the labels attached to crowd behaviours are often subjective and/or value-laden. Here, we present an alternative approach, loosely based on the statechart formalism from computer science. This uses relatively "agnostic" labels, which means that we do not prescribe the behaviour of an individual, but provide a context within which an individual might behave. This naturally describes the time-series evolution of a crowd as "threads" of states, and allows for the dynamic handling of an arbitrary number of "sub-crowds".

A dynamic state-based model of crowds

TL;DR

Addressing the need to describe time-evolving crowd behavior beyond static typologies, the paper introduces a dynamic state-based model inspired by statecharts, organized into Assembly, Mode, Structure, and Dispersal phases. It encodes crowd evolution as threads of states, enabling sub-crowd emergence and cross-phase transitions without assuming underlying psychology, and it notes extensions such as transition weights and persistent tags for memory. The framework is designed to be descriptive and integrable with existing simulation approaches for both normal and incident crowds, improving the ability to report on real events and to study safety implications. The approach offers a flexible, extensible standard for capturing macroscopic crowd dynamics and their time-varying configurations across diverse environments.

Abstract

We consider the problem of categorizing and describing the dynamic properties and behaviours of crowds over time. Previous work has tended to focus on a relatively static "typology"-based approach, which does not account for the fact that crowds can change, often quite rapidly. Moreover, the labels attached to crowd behaviours are often subjective and/or value-laden. Here, we present an alternative approach, loosely based on the statechart formalism from computer science. This uses relatively "agnostic" labels, which means that we do not prescribe the behaviour of an individual, but provide a context within which an individual might behave. This naturally describes the time-series evolution of a crowd as "threads" of states, and allows for the dynamic handling of an arbitrary number of "sub-crowds".
Paper Structure (10 sections, 3 figures)

This paper contains 10 sections, 3 figures.

Table of Contents

  1. Introduction
  2. Previous work
  3. Our dynamic state-based model
  4. Assembly phase
  5. Mode phase
  6. Structure phase
  7. Dispersal phase
  8. Possible extensions
  9. Example application
  10. Summary and further work

Figures (3)

  • Figure 1: Dynamic state-based crowd model, showing Assembly (yellow), Mode (green), Structure (blue) and Dispersal (grey) phases.
  • Figure 2: Evolution of an imagined crowd event. (a) Crowd gathers in square, with two main groups - (1) the mobile "streams" and (2) those already assembled. (b) The audience has assembled. (c) A sub-group (3) begins to demonstrate. (d) The police respond with a cordon. (e) This creates conflict, and some members of the crowd leave in response. (f) The remainder of the crowd is dispersed by the police.
  • Figure 3: Mapping of the event onto our model. Threads (1-3) represent the sub-crowds identified in our analysis.