I N T R O D U C T I
O N T O A P P L I C A T I O N
S
An
industrial practitioner doesn’t care about the
elegance or the complexity of a technology, but rather
about the cost-effectiveness of the solution introduced. We
therefore need to find applications where using multi-agent
systems would yield efficient solutions:
Distributed
application
This comes trivially since multiagent systems are
distributed artificial intelligence (DAI). We however
specify that by distributed applications, we mean
applications where data and information are processed at
geographical different locations in the environment
(spatial distribution) at different times (temporal
distribution). The information and the data are structured
in different groups whose access and use requires specific
capacities (semantic and functional distribution).
Complex
application
Applications that are too complex in the sense that they
are too “large” and require too many
computations if they were to be solved by a single
centralised system. These limitations come from the current
level of hardware and software technologies. For such
applications we divide the solution process across multiple
parallel entities capable of intelligent co-ordination.
Examples:
Simulations:
Simulations
need to represent many different entities and are often
decentralised. Each entity can be represented by an agent,
and all agents interact or work together. There are many
different possible simulations:
•
Swarm systems representing
biological systems where every agent is an ant.
•
Modelling
transportation systems (traffic) for optimisation where
every agent is a vehicle.
•
Social
interaction, simulation of complex social phenomena.
Here is a simulation of crowd
control (.mov
file).
Electronic
Commerce: Agents
are in charge of buying and selling goods for their users.
Medicine:
Agents
can help doctors diagnose patients based on information on
symptoms.
Industry:
We can
also apply Agents to industrial problems (see Application
in Industry) .
•
Optimisation
of industrial manufacturing and production processes.
•
Analysis
of business processes within or between companies.
Source: Multi
agent systems: A Modern Approach to Distributed Artificial
Intelligence,
Chapter 9, Pages 377-409.