Crowd Modeling and Simulation: towards 3D Visualization

1
Crowd Modeling and Simulationtowards 3D
Visualization
7th International Conference onDesign Decision
Support Systems in Architecture and Urban
Planning Sint Michielsgestel 3-5 July 2004

• Stefania Bandini, Sara Manzoni, Giuseppe Vizzari
• Knowledge Engineering Lab
• Department of Computer Science, Systems
• and Communication
• University of Milano-Bicocca - ITALY

2
The intelligence of collective behavior
Different behaviors (movements, speed, reacting
times for each single individual
entity) Example 350 elements Path management
(by fields and sensors) Collisions management
3
The intelligence of collective behavior
Simulation of crowding dynamics single agents
influence the formation of groups Crowd is not
uniform because of the behavior rules of single
agents (shape, reactivity, ...) groups fusion
4
Crowd Modelling and Simulation
Crowd modelling and simulation allows to obtain
realistic behaviours of human-like agents in a
model of the environment It is thus possible to
check architectural designs in specific
situations This kind of instrument allows a
design-simulate-evaluate cycle for design and
planning processes It is a paradigmatic example
for studying the behavior of complex systems
5
Main Approaches
Analytical May handle very large simulation
scenarios Entities as mere numbers No strong
notion of space
Cellular Automata based May handle a large number
of entities Explicit representation of the
environment Entities are homogeneous (they are
conceived as particular states of
cells) Extensions to the basic model are often
required (e.g. action-at-a-distance) Complex
behaviours require a very large cell state and
transition rule

• Multi-Agent Systems based
• May handle a smaller number of entities
• Entities are clearly separated by the
  environment
• Entities may be heterogeneous
• Only a few approaches and models provide a
  representation of the environment

6
Two main computational approaches

• CELLULAR AUTOMATA

MULTI-AGENTS SYSTEMS
7
Cellular Automata

• Introduced in the late 1940s by John von Neumann
  and Stanislaw Ulam.
• CA discrete dynamical systems often described as
  a counterpart to partial differential equations
  (continuous). Discrete space, time and
  properties have only a finite, countable number
  of states. Behavior a system is by interaction of
  cells following rules
• Not to describe a complex system with complex
  equations, but let the complexity emerge by
  interaction of simple individuals following
  simple rules
• In the late 1980s the interest on CA arose
  again, as powerful computers became widely
  available. Today a set of accepted applications
  in simulation of dynamical systems are available
• An example of "macroscopic" dynamics resulting
  from local interaction is "the wave" in a - say
  soccer-stadium
• Each person reacts only on the "state" of his
  neighbor(s). If they stand up, he will stand up
  too, and after a short while, he sits down again
• Local interaction leads to global dynamics

8
Cellular Automata

• A regular n-dimensional lattice (n is in most
  cases of one or two dimensions), where each cell
  of this lattice has a discrete state
• CA develop in space and time
• the number of states of each cell is finite
• the states of each cell are discrete
• all cells are identical (uniformity)
• no action-at-a-distance is allowed
• the future state of each cell depends only of the
  current state of the cell and the states of the
  cells in the neighborhood
• the development of each cell is defined by rules

9
Crowd dynamic models comparison
10
CA-based crowd models

• Environment ? bidimensional lattice of cells, as
  an abstraction of the actual environmental
  structure
• Pedestrian ? specific state of a cell (e.g.
  occupied, empty)
• Movement ? generated thanks to the transition
  rule (i.e. an occupied cell becomes empty and an
  adjacent one, which was previously vacant,
  becomes occupied)

• Choice of destination cell in a transition
  generally includes information which is not
  provided by basic CAs
• Benefit-Cost/Gradient predefined information
  related to cell desirability
• Magnetic Force/Social Force model the effect of
  presence of other agents in the environment
  (attraction/repulsion of crowds)

11
Multi Agent Systems

• Multi-agent systems (MAS) represent a new
  developing area of research (from Artificial
  Intelligence) to be applied as a technology for
  solving problems in an increasingly wide range of
  complex applications
• MAS are inspired by models from biology
  (ecosystems) and economics (markets). They
  represent a new way of analyzing, designing,
  simulating, and implementing complex software
  systems
• Agent theory concerns the definition of agents
  and Multi-agent systems, properties,
  architectures, communication, cooperation and
  coordination capabilities
• The practical side concerns the agent languages
  and platforms for programming and experimenting
  with agents

12
Multi Agent Systems

• A MAS consists of a set of agents
• defined by their behaviors and characteristic
  parameters
• located in an environment where interactions
  occur
• heterogeneous and asynchronous
• MAS models do not take into account the spatial
  structure of agent environment
• This happens despite of the fact that
• Recent results in complexity science suggest that
  the topology of agent interaction is critical to
  the nature of the emergent behavior of the MAS
• A large class of problems is characterized by
  unavoidable spatial features
• Several domains deal with
• space itself (e.g. geographical location)
• a model of it (e.g. information flow in an
  organizational structure)

13
MAS and CA

• CA model offers a computational framework to
  model and simulate natural and artificial
  phenomena involving space
• CA as a kind of MAS
• Spatial structure of agent environment is
  explicit, uniform and regular
• Agent representation implicit in cell state
  description, do not move homogeneous and dense
• Agent behavior is synchronous
• E.g. Applied to analyze urban system dynamics and
  pedestrian activity
• Cellular Space ? dynamics of the urban
  infrastructure
• MAS ? dynamics of the interacting entities
  populating this infrastructure

14
Situated Cellular Agents (SCA)

• A formal and computational framework where to
  describe, represent and simulate complex systems
  that require
• spatial features to be explicitly considered
• different forms of interaction to be integrated
• SCA relaxes constraints on uniformity, locality
  and closure of CA

Open systems can be modeled
15
Situated Cellular Agents (SCA)

• A MAS based modeling approach to represent
• Heterogeneous agents ? multiple types
• Agent environment ? agents are situated in a
  graph structure influencing their behaviors
• Spatially dependant interaction mechanisms ?
• Local and synchronous (i.e. reaction)
• At-a-distance and asynchronous (i.e.
  diffusion-perception-action)

16
Agent Type
An agent type A? is denoted by

• ?? the set of states that agents of type ? can
  assume
• Perception? is a function associating to each
  agent state the vector of pairs representing the
  receptiveness coefficient and the agent
  sensibility threshold to a certain field
• Action? denotes the set of actions that agents
  of type ? can perform

Agent Structured Environment

• Space set P of sites arranged in a network
• Each site p?P is defined by ltap ,Fp ,Ppgt where
• ap? A ? ? agent situated in p
• Fp? F set of fields active in p
• Pp? P set of sites adjacent to p

17
Focus on Field-based Interaction
Emission an agent emits a signal specifying its
characteristic parameters (intensity, content,
diffusion function) Diffusion emitted signals
spread throughout the spatial structure of the
environment Perception Agent state determines
receptiveness and sensitivity Receptiveness
modulates field intensity (amplify or
attenuate) Sensitivity filters not perceivable
signals (low intensity)
18
Primitives for agent behaviour

• Intra-agent actions
• trigger specifies that an agent must change its
  state upon perception of a specific field
• transport defines a rule for agent movement
  (i.e. conditions, destination)
• Inter-agent actions
• emit allows an agent to diffuse a field on the
  site it is placed on
• react specifies that an agent must change its
  state upon perception of a specific field

19
SCA and Crowd Modelling

• Pedestrians ? agents
• Environment ? graph, as an abstraction of the
  actual environmental structure
• Movement ? generated thanks to the field
  diffusion-perception-action mechanism
• Sources of signals (fields) objects, gateways,
  but also agents
• Agents are sensitive to these signals and can be
  attracted/repelled by them
• Possible superposition of different such effects
  (amplification/contrast)

20
From 2D to 3D

• Java based bidimensional simulator implementing
  SCA elements
• Exported log of the simulation including
• Definition of the spatial structure
• System dynamics
• MaxScript that allows 3D Studio Max to generate
  an animation representing the simulated scenario

Avatar001001001004003000_at_ Avatar002002001
003005000_at_ Avatar001002010003002000_at_ Avat
ar002001001003004000_at_ Space001001001001
006000_at_ Space001002001002005000_at_ Space001
003001004005000_at_ Space001004001004004000
_at_ Space001005001003004000_at_ Space001006001
003002000_at_ Space001007001004002000_at_ Space
001008001005002000_at_
21
Sample Application Crowd in a Lecture Hall
One exit
22
Sample Application Crowd in a Lecture Hall
Two exits
23
3D Visualization simple
24
3D Visualization more complex
The image of Scala Square appears courtesy of
GeoSim Systems
25
From 2D to 3D Visualization
The image of Scala Square appears courtesy of
GeoSim Systems
26
Current and Future Works

• Semi-automatic generation of spatial abstraction
  of the environment
• Full 3D solution
• Integration of SCA model concepts into a 3D
  engine
• Possibility to dynamically change the viewpoint
  during the simulation
• Possibility to interact with/during the
  simulation
• Development of other supporting tools
• To visually define field sources, starting
  intensity, diffusion functions, and so forth
• To define agent behaviours
• To define other elements of the simulation
  scenario (types, number of agents, starting
  placement, and so on)
• Integration of psychological/sociological
  information to guide agent behaviour

27
Interactive SCA

• VIRTUAL MUSEUM the user avatar is guided in the
  museum
• - If the avatar is far, the guide calls him and
  waits for him
• - The guide allows the avatar to make decisions
  on current interest (no pre-planned visit)
• Other avatars are in the museum (collision
  management)
• - The avatar can call the guide and ask for
  information

28
Stefania Bandini Dipartimento di Informatica,
Sistemistica e Comunicazione Università degli
Studi di Milano-Bicocca bandini_at_disco.unimib.it

• Thank you!