Simulating Dynamical Features of Escape Panic

Download Report

Transcript Simulating Dynamical Features of Escape Panic

Simulating Dynamical Features
of Escape Panic
Dirk Helbing, Illés Farkas, and
Tamás Vicsek
Alex Turek
Stampedes
•
•
•
•
•
People trying to move faster than normal
Physical interactions between people
Uncoordinated passing of bottlenecks
Arching/clogging at exits
Dangerous pressures within jammed
crowd (up to 4,450 N/m)
Stampedes (cont’d)
•
•
•
Injured/fallen people turn into obstacles
Tendency of “mass behavior,” i.e. herd
instinct
Alternative exits are overlooked or
underused
Studying Stampedes
• Mostly social psychology (mentality of
herd behavior)
• Helbing et al. attempted to model them
using self-driven particle systems
• Combine socio-psychological and physical
forces
A Single Particle
Affected by multiple forces:
Desired Velocity and associated acceleration
Interaction forces
Desired Velocity Force
Helbing et al. specified these parameters:
vi0 (Desired velocity)
•
•
•
•
0.6 m/s – Relaxed
1.0 m/s – Normal
1.5 m/s – Nervous
>1.5 m/s – Panic (transition to uncoordinated
bottleneck behavior)
Interaction Forces
• Psychological desire
to have room
between you and
another person
• Psychological desire
to not be too close to
the walls
F
F
Interaction Forces
• Body force –
counteracting body
compression
• Sliding friction force –
counteracting
movement tangential
to something you’re in
contact with
F
F
F
Measured Effects:
Transition to uncoordination due to clogging
Arch-like bunching, with avalanche effect
when arches break (demo)
Measured Effects:
•
•
•
•
Faster-is-Slower Effect
Attempting to have too high a vi0
High interpersonal friction becoming the
dominant force at a bottleneck
Buildup of forces causing extremely high
pressures, and injuries (demo)
Asymmetrical columns can improve
outflow and prevent buildup of fatal
pressures (demo)
Measured Effects:
Faster-is-Slower Effect
• Jamming can occur
at widening of
escape routes (demo)
Measured Effects:
Mass Behavior
• Individualism vs. Herd behavior
• Each pedestrian may either select individual
direction, follow avg. direction of his neighbors in
a certain radius, or a mixture of both, weighted
by panic parameter pi. (demo)
Desired direction at time t
Measured Effects:
Mass Behavior
• Both have drawbacks:
– All individualistic means no one will learn from anyone
else (demo)
– All herding means no exploration for other exits
(demo)
Measured Data
Leaving time vs. Desired
velocity
Measured Data
Leaving time for 80 people, given
different combinations of individualistic
and herding behavior
Questions?
Cited Paper:
Helbing, Farkas, Vicsek.
“Simulating Dynamical Features of Escape Panic”. 2000
http://www.eecs.ucf.edu/~lboloni/Teaching/EEL6938_2007/papers/H
elbing-EscapePanic.pdf
Presentation by Alex Turek