What comes to mind when you hear the word agent? An agent for a professional baseball player or an agent for a movie star? What comes to mind would vary from person to person. What we mean by agent here is someone who will carry out a certain action. However, that someone is not a real person, rather it is an extracted model of someone who would autonomously act a certain way.
A multi-agent system is a system to solve problems by preparing multiple agents.Let us consider the implication of multiple agents. What happens? Maybe the agents will get in the way of one another. Or, perhaps they would cooperate. Or maybe they would not care about one another and carry on whatever they are doing. It is not easy to predict what exactly would happen, but it is possible to imagine there will be some interaction. In a multi-agent system, you strategically place agents who can carry out certain actions to interact with one another in order to accomplish a particular goal.
It is not easy to predict the outcome when vast numbers of agents interact. That is why we use a computer simulation to examine this in our lab.
A Practical Example
At a Factory
Factories that produce goods can be called a multi-agent system. The people and machines which do the work are the agents, and when they all come together in a certain way they are able to produce goods.
Spatial Prisoner’s Dilemma
This is a model in which multiple agents that are spread apart in space such as on a network play the game called the Prisoner’s Dilemma.
Self-Healing Network Model
This is a model in which nodes within a network fix each other to maintain stability.