Research in
Agent-Based
Modeling
with
Mike Long
Contact Information
melsch@RIT.edu

(585) 475-5158
Carlson 76-3131
 

Agent Based Modeling of Crowd Dynamics


FacultyKevin Bierre

Kevin Kochersberger
(Virginia Tech)

Mike Long

Harvey Rhody
StudentAlex Glade

Holly Zelnio

Agents on Computer Resource Landscape


Crowds develop into essentially two entities – static and dynamic crowds.  Static crowds are found as audiences at performances, concerts, speeches, and rallies.  Dynamic crowds result from marches, when the crowd is following a movement of an event, or visiting various features such as in an outdoor festival or a trade show.  Nevertheless, all static crowds eventually become dynamic upon egress from the event.  However, static crowds can also become dynamic under conditions of physical or external stress such as fire, fights, shootings, etc.  They can also become dynamic with physiological motivation such as protest marches following a speech.
 
Regardless of origin, any crowd can develop into a dynamic panic or malicious entity.  Consequently does this present an opportunity for remote sensing in the monitoring, prediction, and minimization of deleterious impact of large crowds under adverse conditions?  Obviously, the ‘outlier-trigger’ event, such as a fire or bomb cannot be predicted, but can its ultimate impact on the crowed be minimized? Specifically,

  • How does an image of a large crowd appear as it develops from a static, peaceful entity, to a dynamic crowd, to the panic situation?  Is there predictability in this transition?
  • Are there physical and security conditions that exist to minimize a potential panic situation?  Where are the optimal geographical locations to maximized there effectiveness?  Do they vary by event or are there general rules?
  • Is there an opportunity for remote sensing coupled with agent-based modeling for monitoring, short-term prediction and impact minimization of dynamic crowds?