INTELLIGENT CONTROL OF MATERIAL HANDLING SYSTEMS - Environmentally Conscious Materials Handling - page 71

Agent

Perceive Act

Environment

Figure 3.3 Classic agent illustration

There might not be some clear distinction between intelligent or nonintel-
ligent agents, but intelligence is usually combined with some sort of learning
mechanism, and Wooldridge and Jennings add these properties to the classifica-
tion of an intelligent agent (Wooldridge and Jennings 1995):
• Social . Intelligent agents can interact with other agents and systems, includ-
ing humans.
• Proactive . Agents not only respond to their environment, but they can
exhibit goal-directed behavior on their own initiative.
• Reactive . Agents are capable of perceiving changes in their environment,
and react upon that.
Other characteristics, such as adaptability, mobility, and rationality, just to
mention a few, have been related with the agent model, but is it commonly
accepted the four concepts just described (autonomous, social, proactive, and
reactive) are the main characteristics of an agent.

6 MULTIAGENT SYSTEMS

Single-agent systems are very close to traditional centralized monolithic systems,
but the real strength of agents is revealed when multiple agents interact. Therefore
a multiagent system (MAS) is defined as follows:

A multiagent system is a collection of interacting agents (Bussmann et al. 2004).

It is important to state that a MAS is far more than just a collection of agents, as
it also encompasses the emergent behavior, which spawns from the organization
of agents and their interaction under the influence of the environment in which
they are situated. Silva and Demazeau (2002) proposed the Vowels formalism to
model a multiagent system:
MAS = A (Agents) + E (Environment) + I (Interactions) + O (Organization)

• Agents . Agents are the classic entities to consider, when developing a MAS
system, as agents are determined to be the local actors carrying out the
tasks of the system. Agents are the key concept that comprises the system,
which follows the characteristics of single agents.