Programmists - Intelligence.

          ИСКУСТВЕННЫЙ ИНТЕЛЕКТ           
              /волчьи мысли/              
      ═══════════════════════════════     
                                          
                                          
                                          
(С) WOLF/RAGE                             
(W) SERGEY STURM/LGN                      
---------------------                     
                                          
                                          
     Про искуственный интелект уже сказано
довольно много. Особенно выделяется в этом
ZX-FORMAT 5 & 6. Но  оба метода предложен-
ные  там, обладают  некоторыми недостатка-
ми. Волновая трассировка работает довольно
медленно /это  признает сам В.Медноногов/.
А  метод расчета лучшего хода по оценочной
формуле годится практически толькоfor lo-
gical games such as chess and reversi. But what
- the same thing to do if you are writing dynamic
a game where you need quick intelligence? Yes
several methods...              
     First, let's define:     
We are not making a thinking character, but a persona
nazh, which only seems to be thinking. K
for example, if you saw Monstrland Level 4
(and you haven’t seen it), then the characters are there
They hang around you, periodically shooting.
If you try to approach him, he will
dit from direct combat. And if you rush to
wall, then only then will he rush to attack.
And on top of everything else, they make it quite difficult
get off. In short, they create the impression
"thinking".  And this is achieved quite
simple methods.                        
     The most necessary thing is to determine
board to point. Let's say the coordinates
your character xp, lvl. Coordinates to where
he has to go: ho, wo. Encoding direction
expressed by number:                              
                                          
                                          
                                          
If 4, then go  ^ /         
nowhere. We already do so  |   /         
уже здесь.                |  /           
                         0 1 2            
                     {---3 4 5---}        
                         6 7 8            
                        /  |             
                       /   |             
                      /    V             
                                          
                                          
Алгоритм вычисления направления           
--------------------------------          
                                          
1. р=хр-хо                                
2. Если р<128, то sx=-1  ; xр-xо>0? sx=#ff
3. Если р>128, тоsx=1  ; хр-хо<0? sx=1   
4. Если р=0, то sx=0  ; хр-хо=0? sx=0     
5. р=ур-уо                                
6. Если р<128, то sy=-1  ; yр-yо>0? sy=#ff
7. Если р>128, то sy=1  ; ур-уо<0? sy=1   
8. Если р=0, то sy=0  ; ур-уо=0? sy=0     
Выборка                                   
--------                                  
                                          
9. Если sx=-1, то n=[0,3,6]               
10.Если sx=1, то н=[2,5,8]                
11.Если sx=0, то n=[1,4,7]                
12.Если sy=-1, то р=n1 /первое число из n/
13.Если sy=0, то р=n2 /второе число из n/ 
14.Если sy=1, то р=n3 /третье число из n/ 
            
     That's all! After working out this algorithm
ma in r you have a direction code. By the way-
Please note that this algorithm is not optimal.
timized. More or less experienced programmer
the mist has probably already seen where it is possible to
shorten checks, etc.      
     But that's just one thing. Let's say the code is
we have boards. But how to do it
code move? Elementary! I'm in Monsterland
made a table of displacements by coordinates,
where the offset number was the direction code
nia. It should look something like this:     
                    
(-1=#FF)             
     We believe that the 1st element is the displacement
offset in X, the second element is the offset in y.
Now we make a selection from the table.         
                                         
 ;C - direction code           
 ; the TAB table is located in a “round”
                                HEX address
                                          
   LD A,С                                 
   ADD A,A                                
   LD Н, ТАВ /ст. байт адреса таблицы/    
   LD С,A                                 
   LD A,(HL)                              
   INC L                                  
   LD Н,(HL)                              
   LD L,A                                 
                                          
     Теперь  у нас в HL смещения относите-
льно координат. Делаем так: ХР+L ; УР+Н  и
это  будут  наши  новые  координаты. Можно
сделать еще проще и без таблицы. Посмотри-
те вверх. Ведь мы, по-сути, уже  вычисляли
смещения т.е. SX и SY. Достаточно  там  их
сохранить и использовать здесь. Но так де-
лать я вам не советую, иначе  потом  могут
появиться лишние сложности.                                             
                                         
          ARTIFICIAL INTELLIGENCE           
 (PART II)                 
                                
    
     So, we have learned to choose a direction
tion and make a move in this direction.
As you understand, this is clearly not enough
for writing really strong in-
llekta. We need a tool that allows
which controls the direction of travel. Such
there is a tool.  This is the mechanism that carries
the name is "finite state machine". Despite
ease of implementation, it allows you to get
pretty good results.             
     So, what is a finite state machine?
This is a set of states, i.e. actions that
things your object can do. Let's say
you have decided what the character can go to
to the player. If he is heavily armed, he runs away
from him. If the distance is not too great -
Shoe, then he unleashes a hurricane on the player
fire. That's all for now.             
                   
 So we have states:             
    
1. Go to the player                         
2. Run away from him                        
3. Shoot the player                     
                                         
     In the object description we create a cos byte
storage in which the current state is stored
tion of the object at the moment. Next, create
we define the rules for transition from one state to
other.                                   
   THOSE.                                   
 Current state 1. Rules:         
                                         
 If the player is armed, then go to the co-
   standing 2                               
 If the player is close, then go into the state
   3              
 Current state 2.                 
             
 If the player is far away, then go to the state
   1                         
                        
 Current state 3.                   
                                         
 If the player is out of shot range,
   go to state 1                 
                                         
     In short, in assembly language this is
writing is easier than words. For each suck-
When you write a processing program and you-
take them depending on the status byte
ny. Each such program must include
transition rules, otherwise the character will never
will not come out of this state.             
What we already have is enough for up to
loosely “smart” intelligence. What I'm talking about
talking further is considered aerobatics
in programming intelligence. I'll go-
talk about emotions. Imagine that
the choice of state is not strictly controlled
established rules of transition, and emotions, action
acting on the principles of implicit logic. That is, co-
the “retreat” state is caused by a combination
tion of two emotions: “panic” and “fear”. Feel-
"hunger" forces you to look for food/food
may well be the main character or 
same object/.  Emotion "caution"
indicates that one must move carefully,
using shelters, etc. This is really
cool! You can come up with a bunch of different ones
emotions and assign them different priorities.
But just don't overdo it. Otherwise the movement
your subject's performances will resemble a dance
crazy...                            
 So I told you about my thoughts
in the field of high-speed I.I. I didn't purposely
I didn’t drive a single program, but showed
only the methodology. All implementation methods you
develop it yourself, if, of course, you want to
those After all, my story is very superficial and
You'll have to figure out a lot of things yourself. But in
That's the fun of writing games.
Usually, just a small amount is enough
push, sometimes one word, to be born
IDEA...------------------------------------------