Multi-Agent System

The widespread use of distributed information services will radically alter the way in which both organizations and individuals work. There are many indicators of this coming information revolution.

Though there exists no generally accepted definition, an agent can be thought of as an entity with domain knowledge, goals and actions. An autonomous agent is capable of interacting independently and effectively with its sensors and effectors in order to achieve specified or self-sired goals.
 
Russell and Norvig define an agent as: Anything that can be viewed as perceiving its environment through sensors and acting upon that environment through effectors.
 
The study of autonomous agent behaviors, the way in which an autonomous agent acts, is advancing rapidly as a sub-field of distributed artificial intelligence (DAI). Creating an intelligent autonomous agent to the level of human performance is one of the dreams of AI researchers.
Multi-agent systems deal with the construction of complex systems involving multiple agents and their coordination. It inherits from the distributed AI. Distributed AI (DAI) systems are not capable of emergent contexts and changing problem-solving roles for concerned agents.
 
A multi-agent system (MAS) is a distributed computing system with autonomous interacting intelligent agents that coordinate their actions so as to achieve its goal(s) jointly or competitively. The agents in a multi-agent scenario, may have homogeneous or heterogeneous structures. In a homogeneous MAS, all of the agents have identical structure (goals, domain knowledge, and set of actions). They may differ by way of their sensor input and effector output. In a heterogeneous situation, agents can have different goals, domain knowledge and actions. The agents in such a system may be friendly (benevolent) or may be inhibiting each other (competitive). The agents, situated differently in the environment, receive different sensory inputs and initiates different actions. Heterogeneity imparts much power to the MAS at the cost of complexity.