Robotics Reference
In-Depth Information
list of which squares a particular piece type may be able to move to from
a certain square: a white pawn on the square a2 can move to a3 if that
square is empty, it can also move to a4 if both a3 and a4 are empty, while
if the square b3 is occupied by a black piece then the white pawn on a2
may capture it. All this is taught to human beginners and is easy enough
to program.
How Computers Evaluate a Chess Position
Once a program has the ability to generate a list of all the legal moves
from any position, it needs a way to decide which of these legal moves
is the best one. What it does, in simple terms, is to examine each of the
positions that can arise after its next move and compare the merits and
demerits of these positions, employing an algorithm called an evaluation
function.
2
What this function does is to assign a numerical score to a
Chess position in such a way that a high score indicates a good position
for the program and a low or negative score indicates a bad position. The
program then chooses the move leading to the position with the highest
score.
An evaluation function determines the score for a position by taking
into account various features of the game. The most important feature
in Chess is what is known as
material
. It is clear that the different pieces
each have their own values—Chess players call the total value of their
pieces the material value. One of the simpler heuristics used by Chess
players is that it is normally a good idea to capture one's opponent's pieces
for nothing whenever possible, or for pieces of lesser value. Experience
has shown that the following numerical values are a very reasonable in-
dication of the relative worth of the different pieces, and these were the
values proposed by Shannon:
3
Pawn
=
1
Knight
=
3
Bishop
=
3
Rook (castle)
5
=
Queen
9
=
2
Sometimes called a scoring function.
3
The kings are beyond value because a player who is faced with the certain loss of his king
automatically loses the game.
