1. Distributed Agent-Based Building
Evacuation Simulator
A. Filippoupolitis, E. Gelenbe, D. Gianni,
L. Hey, G. Loukas, S. Timotheou
{gianni, e.gelenbe}@imperial.ac.uk
Intelligent Systems and Networks Group
Imperial College London
2. Presentation Overview
Emergency management
Motivation
Simulated Model
Simulation Framework – Building Evacuation
Simulator (BES)
Integration with a wireless sensor network
Conclusions
3. Building Evacuation Scenario
Scenario characteristics:
• Multi-storey building
• Emergency situation
• Civilians try to evacuate
following the quickest and
safest path to the exit, while
adapting to the events
• Emergency personnel
enter the building trying to
rescue civilians and
extinguish fire
4. Motivation
Decentralised optimisation techniques that will support
actors during dynamic and rapidly changing situations
We want to carry out systematic investigations of such
techniques in largely populated scenarios
A framework is needed, that allows:
1. Reproducibility of experiments
2. Extendibility to diverse scenarios
3. Distributed operation (for largely populated scenarios)
5. Simulated Model
The model includes:
• A world Model, which represents the physical space
inside the building and its status
• One or more hazard agents, which affect the status of
the world
• A population of human agents, which move and
cooperate inside the physical world according to
personal characteristics
6. World Model
A graph models the physical space:
• The nodes of the graph represent “Points of
Interest”
• Each edge represents a physical path
between two nodes
Graph elements are enriched with a set of
attributes that represent the status of the
world
7. Example of World Model
Each node has a queue of human agents, attributes for fire, x and y
coordinates in the space, type of node (e.g. door, stairs), etc.
Each edge has: list of human agents traversing it, attributes for fire,
etc.
8. Types of Agents
Three type of agents
• Resource Manager, which manages the
access to the world nodes
• Human agents, which move inside the
physical world
• Hazard agents, which affect the status of the
world
9. Resource Manager (RM)
RM is in charge of:
• Coordinating the access to the nodes
• Providing world updates for each agent
RM is defined by a simple wait
event/process event logic that proceeds
until the simulation ends
10. Human Agents (HAs)
HAs are characterised by:
A personal view of the world
One or more goals (including the decision
models on how to achieve them)
Motion model
Health model
11. Going Distributed
Why? The amount of computational resources
grows at least as polynomial function of the
number of simulated agents
Two major modelling issues to face:
• Model partitioning
• Model adaptation to the distributed environment
12. Model Partitioning
We follow three guidelines:
• Exploiting the intrinsic parallelism of independent
physical subsystem
• Meeting local memory constraints
• Minimising the network workload
The simulated world is allocated on independent
single area simulators (floor or stairs) running
on a separate host
13. Model Adaptation
The performance of the simulator are affected
by the amount of data exchanged
Reduce such data by:
• Locally store “constant” data
• Move only individual knowledge
Agents also interact with local world only
Locally, condensed representation of the remote
world (GPoI)
14. SimJADE
SimJADE is:
• A Java framework for Agent-based M&S
• JADE-based, thus FIPA compliant
It offers a formulation of discrete event
simulation systems in terms of MAS through its
components
It also provides a uniform interface for MAS and
Agent-based M&S, easing therefore the
development of such simulators
15. SimJADE components
It is defined through:
A simulation ontology
• Simulation time, simulation services
A simulation agent society
• Simulation engine (local/distributed), which orchestrates the
simulation
• Simulation entities, which incorporate the logic
An interaction protocol between the agents, implemented
by a set of behaviours and simulation event handlers
16. • A virtual hazard (fire, gas, etc.) spreads
inside the physical world
• A real Wireless Sensor Network test-bed
monitors the spreading of the hazard
• Each sensor is assigned to a vertex on the
graph of the emergency response simulator
(e.g. like a room's smoke detector)
• We use light from LEDs to represent the
hazard within the virtual building
• The hazard agent controls the hazard
spreading and the intensity of the respective
LEDs, providing input to the sensors
regarding the intensity of the hazard
Wireless Sensors Test Bed Integration (1)
17. • Two different representations of the virtual hazard spreading (sensed , actual):
- The Building Evacuation Simulator (BES) connects to the wireless sensor
network, processes the sensor readings and updates the sensed
representation of hazard spreading
- The actual data from the fire simulator are also processed by the BES in
order to update the actual representation of hazard spreading
• Effects of hazard spreading:
- Over simulated time the paths become more hazardous and “slower” to
traverse
- When actors move along an edge with increased degree of danger, their
health level decreases
- Excessive exposure to danger results in a fatality
Wireless Sensors Test Bed Integration (2)
18. Current Building Evacuation Simulator
Floor1
Floor2
Floor3
Floor4
Stairs
A
Stairs
C
Stairs B
Point of Collection
…
19. Current applications
• Adaptive on-line decision
support for building evacuation
• Optimal allocation of rescuers
to injury locations
20. Conclusions
Develop decentralised optimisation techniques that can provide
decision support during an emergency situation
Such techniques require a systematic investigation before being
deployed in real scenarios
Cost and time effective investigations require a software framework
that combines:
• experiment reproducibility
• high level of extendibility
• distributed operation
We presented the Building Evacuation Simulator, a simulation
framework that meets such requirements, and some basic examples
of use