Multi-Agent Systems

• An Introduction to MAS
• Introduction to Applications
  • Applications of MAS
  • Industrial Agents
  • Product Design
  • Planning & Scheduling
• Swarms
• Glossary
• Bibliography

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.