Bentham & Hooker's Classification. along with the merits and demerits of the ...
Articial societies immersed in an Ambient Intelligence Environment
1. Articial societies immersed in an Ambient
Intelligence Environment
Teresa Garcia-Valverde, Emilio Serrano, Juan A. Botia, Antonio
Gomez-Skarmeta and Jose M. Cadenas
{mtgarcia,emilioserra,juanbot,skarmeta,jcadenas}@um.es
University of Murcia
8 de julio de 2009
Teresa Garcia-Valverde Et al. Societies in AmI environments
2. Proposal and motivation
The use of MABS (multi-agent based simulations) to develop
and test AmI (Ambient Intelligence) applications
Applications where intelligent embedded objects into an
environment that is able to recognize and to respond to
dierent individuals
Real tests may be too costly or impractical.
Thousands of tests
Thousands of people
Emergencies
Simulations can be used
Great exibility.
Cheap and very expressive.
Replacing other modes of simulation.
sociology, biology, physics, chemistry, ecology, economy, etc.
Teresa Garcia-Valverde Et al. Societies in AmI environments
3. Application
Used for emergency management in a real building, the
European Center of Business and Innovation (CEEIM) at the
University of Murcia.
http://www.ceeim.es/
Teresa Garcia-Valverde Et al. Societies in AmI environments
4. Application (II)
For the emergency management case is
simulated:
A building
People
Fire
AmI devices:
Sensors
Actuators
AmI applications
The behavior can be observed by
Displays of oors
Displays of stairs
3D History of deaths
Charts
Presentation video UbikSim.mp4
http://ubiksim.sourceforge.net
Teresa Garcia-Valverde Et al. Societies in AmI environments
5. Conguring ubiksim
Simulation can be congured for specic targets
Building:
Number of:
Floors
Corridors
Rooms
Stairs
Width of:
Corridors
Stairs
Doors
etc
Fire:
Combustibility of the
environment
Ignition temperature
People:
Number of people
Distance to perceive:
Fire
Actuators
Other people
eeing
Speed
etc
AmI devices:
Number and position.
Distance to detect
emergencies
etc
Teresa Garcia-Valverde Et al. Societies in AmI environments
6. AmISim architecture
Realism is important
Real features should be integrated
The following parts must be considered in an AmI simulator
Environment model. A model to describe the physical world.
Agent model. A complex multi-agent model that simulates
physical and human behaviour.
Context model. This model gathers, interprets, and stores the
contextual information.
Adaptation model. A model capable of supporting applications
and services which use contextual information.
Teresa Garcia-Valverde Et al. Societies in AmI environments
7. AmISim architecture (II)
Simulation ⇒ Ubiksim ⇒ Agent model and Enviroment model
Real features ⇒ Adaption model and Context model
Teresa Garcia-Valverde Et al. Societies in AmI environments
8. AmISim architecture (III)
AMISIM IS FLEXIBLE!, everything can be changed easily. The user
can:
Adapt it to a specic building (environment model)
Change the type of AmI applications (adaptation model)
Include assumptions and new congurations
etc
Teresa Garcia-Valverde Et al. Societies in AmI environments
9. Examples of AmI applications
Three AmI applications for evacuation have been developed and
tested
Evil
Deactivates the actuators
The agents cannot receive any re alarm.
Simple
Activates all the actuators of the building if a re is detected.
Calculates the direction and distance from the re.
Actuators suggest directions to people.
Stupid
As the simple but without suggesting directions.
Teresa Garcia-Valverde Et al. Societies in AmI environments
10. Examples of results
Results for:
300 tests were performed simulating a building with three
oors and 300 workers (res at random positions).
Average of saved and died agents:
Teresa Garcia-Valverde Et al. Societies in AmI environments
11. Examples of results (II)
Besides the saved and died agents...
we can analyze numerous elements to improve/test our AmI
applications
propagation of re per unit of time
average of time to escape from each oor
deaths following suggestions from an actuator
average of time needed by sensors to detect re
agents using each stairway per unit of time
...without simulations this kind/number of tests could not be done
Teresa Garcia-Valverde Et al. Societies in AmI environments
12. Examples of results (III)
For example:
How many agents have...
...ed because of a re?
...ed because of an actuator?
...followed suggestion from actuators?
Results for the simple application:
Most of the agents start escaping after seeing an actuator
Most of these are saved.
...but only half have been saved following suggestions from
actuators.
Teresa Garcia-Valverde Et al. Societies in AmI environments
13. Conclusions and future works
The MABS are very useful to develop and test AmI
applications
The proposal provides a simulation framework which can test
scenarios that would be impossible in real environments
Shown:
It is being used in real cases.
It is very exible.
Strategies for an adequate evacuation.
How to evaluate those strategies.
Future work will focus on testing our simulator using more real
examples
New strategies for evacuation.
Teresa Garcia-Valverde Et al. Societies in AmI environments
14. Thank you very much for your attention
Contact: {mtgarcia,emilioserra,juanbot,skarmeta,jcadenas}@um.es
More on http://ubiksim.sourceforge.net
Teresa Garcia-Valverde Et al. Societies in AmI environments