A dynamic state-based model of crowds

• DOI: 10.1016/j.ssci.2024.106522
• Publication Date: 2024-04-06T21:37:52Z
• Authors: Martyn Amos; Paul Gainer; Steve Gwynne; Anne Templeton
• Keywords: FOS: Computer and information sciences; Computer Science - Multiagent Systems; Multiagent Systems (cs.MA)

We consider the problem of categorising, describing and generating dynamic properties behaviours crowds over time. Previous work has tended to focus on a relatively static "typology"-based approach, which does not account for fact that can change, often quite rapidly. Moreover, labels attached crowd are subjective and/or value-laden. Here, we present an alternative approach uses "agnostic" labels. This means do prescribe behaviour individual, but provide context within individual might behave. naturally describes time-series evolution crowd, allows handling arbitrary number "sub-crowds". Apart from its descriptive power (capturing, in standardised manner, descriptions known events), our model may also be used generatively produce plausible patterns dynamics as component machine learning-based approaches investigating interventions.