I.1.4.4. Structure of the agents
The work of the artificial intelligence is to conceive
the program of the agent which implements the function of the
agent and then that function projects perceptions on the actions. It is
supposed that the- program will be carried out on a certain type of computer
equipped with the physical sensors and effector
(architecture).
Agent = architecture + program
· Programs of the agents
The programs of the agents that we will describe in this work
have the same architecture: they receive current perceptions like entries of
the sensors and put out an action to the effector.
· Algorithm of description in pseudocode
I describe the algorithms in pseudocode which must be familiar
with the users of the languages Java, C++ and Lisp. But I have to point out
some idiosyncrasies:
v Static variables: We use the key word
«static» for saying that a variable receives an initial value the
first time that a function is called and retains this value in the next calls
of the function.
v Functions: The functions and the procedures
have names in capital letters, and the variables have names in italic and tiny.
A call of a function is similar to FN(X).
v The tables start with 1. The first index of
a table is 1 as in the usual mathematical notation and not 0 as in Java and
C++.
v The indentation is significant. The
indentation is used to mark the field of a loop or conditional, as in the
Python language, contrary to the language Java and C++ (which use accodances)
or Pascal and Visual BASIC language (which use end).
| 
