INTELLIGENT CONTROL OF MATERIAL HANDLING SYSTEMS - Environmentally Conscious Materials Handling - page 72

• Environment . The environment is the space in which the agents exists,
moves, and interacts. The space could be virtual, informational and concep-
tual, but typical the environment is represented by a model of the physical
space of the MAS community.
• Interactions . Interactions and communication are evident in MAS, due to
the aspects of distribution in the systems, and originally by Wooldridge and
Jennings as the social ability of an agent (Wooldridge and Jennings 1995).
Interactions in the MAS community could take many forms; negotiation,
collaboration, coordination, queries, or generally any kind of informa-
tion exchange between agents. Interactions could be formed as abstract
speech-act messages, or in other models as simple natural forces that influ-
ence other agents.
• Organization . Similar to humans, agents can benefit from being organized,
either explicitly defined in classic organizational structures, or the organiza-
tion could emerge from simple interactions among the agents. Organization
often serves the purpose of grouping agents with similar or related actions
or behaviors. Organizations can be helpful to support agents in planning,
performing actions, requesting information, or realize global goals of the
agent system.
In the research of multiagent systems, two different perspectives are
dominating — either a micro-level or macro-level perspective. For example,
the system could focus on the micro-level issues, such as the internal of the
individual agents. What is the decision logic of the agent, how will the agent
learn, and how can it be ensured that the agent will act autonomously? In the
macro-level perspective the multiagent research community is more concerned
with the organization of agent, and how the agent will interact and collaborate in
an efficient way. Whereas the micro-level perspective is commonly inspired by
biological systems, such as the human brain and ant colonies (Parunak 1997),
the macro-level perspective analogies are coming from human organizations
and societies (Ferber et al. 2003, Zambonelli et al. 2000).

7 AGENT TYPES

Agents are usually classified as being either reactive or deliberate in their behav-
ior (Bussmann et al. 2004). Coming out of the artificial intelligence community,
a deliberate or cognitive behavior of agents was expected:
• Cognitive or deliberative agents . A cognitive agent is one that owns a
knowledge base and holds a model of the current environment as it has
perceived it, but it will not act only on a search in the knowledge base. It
has planning capabilities, so it proactively can adjust its actions according
to its goals, even though the perceived input is not covered by the know-
ledge base.