04-05-2012, 05:12 PM
What’s the Name of the Game Formal Specification of Artificial Intelligence Games
Artificial Intelligence for Computer Games 4.pdf (Size: 324.62 KB / Downloads: 37)
Introduction
Game playing is one of the oldest areas of research in Artificial Intelligence.
The first studied games were board games, like chess. In the past years, research
evolved to cover multi-agent games with sophisticated agent interaction,
including simulation of human behaviour.
According to Russel and Norvig [22], an agent is anything that can be
viewed as perceiving its environment through sensors and acting upon that
environment through effectors, or actions. In multi-agent artificial intelligence
games, a competition is established among agents, which are usually restricted
to a small number of actions.
Related Work
There is a great number of available commercial games which use artificial
intelligence techniques to specify the behaviour of bots [26]. Some of them have
bots with a fixed behaviour, designed to fight against human players. Most
first-person shooter (FPS) games are included in this category. FPS games are
very popular 3D action games where the user moves through different scenes
and collect weapons to destroy enemies. One example is Doom II [19]. We
are interested in another kind of games: the ones in which it is possible for
the users to program the behaviour of the bots. They are known as games
with extensible AI [29].
Abstract State Machines
In this section, we present a brief introduction to the formalism of Abstract
State Machines, concentrating in the sequential model. Agents and concurrent
execution will be addressed later. A more formal and complete presentation
can be obtained in [3,8,27].