Information Technology Reference
In-Depth Information
The first problem results in the research of computer vision and natural
language processing. The second problem induces the research on knowledge
representation, autonomic reasoning and planning. In the real world, these
problems are a bit difficult since the representation is very complex. So, the
deliberative agents have its limitation in dynamic environment. Without
necessary knowledge and resource, it is difficult to add the new information and
knowledge about environment into existent model.
Agent
Sensor
Information
Fusion
Internal State
Knowledge
Base
Plan
Environment
Goal
Action
Effector
Fig. 14.4. the architecture of deliberative agent
Deliberative agent is a kind of active software. Different with concrete
domain knowledge, it has the knowledge representation, problem solution
representation, and environment representation and communication protocol.
According to the way of thinking, deliberative agents can be divided into abstract
thinking agents and visual thinking agents. Abstract thinking agents have abstract
concepts. They think through the symbol information processing. However,
visual thinking agents think using the visual material. It adapts the connection
theory of nerve mechanism.
Figure 14.4 give the architecture of deliberative agent. Agent gets outer
information through the sensors. After then, the agents syncretize the information
according to the inner states and generate the description of current state. They
create the plan with the support of knowledge base then. After then, agents can
form a series of actions which is taken by actors to affect the environment.
