Information Technology Reference
In-Depth Information
•
For the calculation of values, the function
f
1
, which is responsible for the realization
of the agent
A
g
1
algorithm, needs internal parameters of that agent (
x
1
), global
data (
x
0
), and internal parameters (all or some) (
x
2
) of the agent
A
g
2
. It should be
emphasized that the agent
A
g
1
does not have access (direct) to parameters (
x
2
).
•
For the purpose of achieving the appropriate data, the agent
A
g
1
observes the
behaviour of the agent
A
g
2
. It means that the agent
A
g
1
observes changes in the
environment (global data
x
0
) which result from the action (realization of the action)
of the agent
A
g
2
. On the basis of the observation result, the agent
A
g
1
may define
(estimate) the state of the agent
A
g
2
, in other words the state of (values) its internal
parameters. In order to do that, the agent
A
g
1
must possess some knowledge about
the agent
A
g
2
, and especially some knowledge about the function
f
2
and its effect
on the changes of the state of the environment
X
0
, as well as its influence on the
state of that agent
X
2
(Fig.
2.13
).
The process of defining the values of parameters (
x
2
) may be realized with greater
or lesser precision, depending on specific, practical possibilities. In effect observed
(estimated) data do not have to give precise, complete information about the state
of the agent
A
g
2
, but they should be sufficient for continuing actions by the agent
A
g
1
(for the calculation of the values of the function
f
1
).
•
The agent
A
g
1
possessing essential information (values of parameters
x
1
,
x
0
and
indirectly
x
2
), using (calculating) the function
f
1
may change its state (parameters
x
1
), and the state of the environment (global data- parameters
x
0
). The changes of
the environment state are realized through calling subsequent actions (actions of
the agents
A
g
1
and
A
g
2
). However, the agent
A
g
1
does not have the access to the
internal data of the agent
A
g
2
and it is not capable of effecting directly the change
of its state, though it should be done as a result of the calculation of the function
f
1
.
Nevertheless, it is possible to achieve it indirectly with the use of changes of the
environment state which forces the change of the state of the agent
A
g
2
. The agent
A
g
2
similar to the agent
A
g
1
observes changes in the environment (parameters
x
0
)
resulting from the actions of the agent
A
g
1
, and on the basis of the information
makes changes of its state, in other words modifies parameters
x
2
.
The procedures of gaining autonomy by an agent give greater independence than
that received in the object-oriented approach because it is not engaged directly in the
internal states of another agent. The solution based on the observation process is more
difficult in realization, however, the range of interaction and cooperation between the
agents gives greater possibilities in the field of forming agent systems (multi-agent).
Numerous problems occur such as intentionality, suitability of actions, awareness,
cooperation between the agents, as well as problems of interaction between the
agents and others which remain open. Some of them will be discussed further in
later chapters. The two approaches to the realization of autonomy of cooperating
algorithms (with the use of communication and observation) provide a basis for
distinguishing between the object notion and the agent notion.
Summing up, it may be noted that the source of information for the agent (
A
g
1
)
is the state of its local data (
x
1
), the state of the environment (global data
x
0
) and
the information received as a result of observation of behaviour of other agents (for
