Kecerdasan Buatan dengan Metode ID3 Finite State Machine dalam Turn-Based Tactics Game

Abstract

As the world of video game grown and advances there are even
more those who likes and play them as well. One way to increase
immersion within video games is by implementing AI (Artificial
Intelligence). AI is an ability within a program or machine that
can grant said machine/program a power of “thought” so that
they can do a command without unnecessary human inputs. And
AI had grown constantly significant since the day it was
conceived.
However one of AI’s greatest weakness is if implemented too
simplistic within it’s execution, then when someone manage to
decode the AI’s pattern, then said player can use this knowledge
for their advantage. And as a consequence, costed the game
immersion since the game is played not the way it was intended to
or become way too easy. To solve said problem, requires a
flexible AI that can adapt to the gameplay. Using ID3 decision
tree, AI are expected to be able to using the variable within a
Turn-Based Strategy Game to use different options according to
the data it earn. And hopefully with little improvements, this
method can be used to differing video game genres as well.
The test result proven that while ID3 does give some influence for
the AI decision making, during the test run for Turn-Based Game
shown that the Adaptive AI that always chooses the best option if
played the same had a WDL overalls (Win-Draw-Lose) of 1:2:0,
while Adaptive AI with the ID3 formula and dynamic decision
making had a more balanced results of 1:1:1 and random AI
always ended at a disadvantage with 0:0:3. These tests proved
that Adaptive AI provide the most challenge to the player doesn’t
have any influence for the early parts of the game. And that the
ID3 impact were non-existent during the early game but
improving as the game goes on and the data table more varied.