Transcript VR-DIS

Towards A Multi-Agent
System for Network Decision
Analysis
Jan Dijkstra
Agenda
1.
3.
4.
5.
6.
Introduction of the Model
Essentials of Cellular Automata
Agent Characteristics
Multi Agent Simulation Models
Towards the Framework
Introduction of the
Model
• Architects and urban planners are often faced
with the problem to assess how their design
or planning decisions will affect the behavior
of individuals.
• One way of addressing this problem is the
use of models simulating the navigation of
users in buildings and urban environments.
A Multi-Agent System
based on Cellular Automata
Essentials of Cellular
Automata
 Cellular automata are discrete dynamical
systems whose behavior is completely
specified in terms of a local relation
Cellular automata are characterized by the
following features:
• Cell
• Grid
• State
• Time
Cellular Automata Model of
Traffic Flow
Agent Characteristics
Agent Definitions
Agents are computational systems that inhibit
some complex dynamic environment, sense and act
autonomously in this environment, and by doing so
realize a set of goals or tasks for which they are
designed (Maes).
An autonomous agent is a system situated within
and part of an environment that senses that
environment and acts on it, over time, in pursuit of
its own agenda (Franklin & Graesser).
Agent Properties
• Autonomy
- agents have some control over their actions and
internal state
• Social ability
- agents interact with other agents
• Reactivity
- agents perceive their environment and respond to
changes in it
• Pro-activeness
- agents exhibit goal-directed behavior by acting on
their own initiative
• ? Mentalistic capabilities
- knowledge, belief, intention, emotion
Agent Architecture
Perception
Action
Production
System
Effectors
Sensors
State
Multi Agent
Simulation Models
Offers the promise of simulating autonomous
agents and the interaction between them.
behaviors evolve dynamically
during the simulation
Evolution capabilities:
• evolution of the agent’s environment
• evolution of the agent’s behavior during the simulation
• anticipated behavior
• unplanned behavior
Towards the Framework
Cellular
Automata
Artificial Intelligence
Distributed
Artificial
Intelligence
Multi Agent
Simulation Models
Motivation
• Develop a system how people move in a
particular environment.
•
•
People are represented by agents.
The cellular automata model is used to simulate
their behavior across the network.
• A simulation system would allow the designer
to assess how its design decisions influence
user movement and hence performance
indicators.
Network Model
The network is the three-dimensional cellular
automata model representation of a state at a
certain time.
transition of a state of a cell
different neighborhoods
von Neumann
r=1
Moore
r=2
r=1
r=2
Agent Model
Conjoint
Measurement
Agent
Virtual Environment
Decision
Support
Agent
Technical
Communication
Actor
Agent n
Actor
Agent 1
Simulation Model
Interface Agency
Virtual
Interaction
Subject
Agent
Intuitive
Communication
User Agent
Define an user-agent as: U = < R | S >,
where:
• R is finite set of role identifiers; {actor, subject}
• S scenario , defined by: S = <B, I, A, F, T>, where:
•
•
•
•
•
B represents the behavior of user-agent i
I represents the intentions of a user-agent i
A represents the activity agenda user user-agent i
F represents the knowledge of information about
the environment, called Facets
T represents the time-budget each user-agent
possesses
The Integration of Cellular Automata
and Multi Agent Technology
Initially, we will realize different graphic
representations of our simulation:
• a network-based view
• a main node-based view
• an actor-based view
network grid and decision points
S1
S2
S3
S4
S5
°
°
°
°
°
E1
E2
°
S16
°
S15
°
S13
main decision point
remaining walkway
section decision
point
°
S17
S14
°
° S6
S18
S20
° S19
°
°
° S7
S12
°
°
S10
section bound
S11
°
E3
°
S9
° S8
main node-based view
links
actual path
actual decision point
actor-based view / network-based view
Simulation Experiment
Design of a simulation experiment of
pedestrian movement.
Considering a T-junction walkway where
pedestrians will be randomly created at one of
the entrances.
Some impressions ...