Asteroids Screenshot

Asteroids is a video arcade game released in November 1979 by Atari Inc.
It was one of the most popular and influential games of the Golden Age of Arcade Games, selling 70,000 arcade cabinets.
Asteroids uses a vector display and a two-dimensional view that wraps around in both screen axes.
The player controls a spaceship in an asteroid field which is periodically traversed by flying saucers.
The object of the game is to shoot and destroy asteroids and saucers while not colliding with either, or being hit by the saucers' counter-fire.

Bachelor thesis

This is, however, something more than just a game, because this application is the accompanying implementation of my bachelor thesis.
In the following the introduction of this thesis:

"An important area of research in artificial intelligence and neuroinformatics
is the artificial or machine learning. The goal is to teach an artificial "lifeform"
independent learning, so it is easier to adapt to specific, varying problems without outside intervention.
As in many other areas of science, there are variants of the artificial
Learning, which are based on natural models. As well as the reinforcement learning, forming the basis of this work. Models are living beings, which learn by interacting with their environment and with the enriched knowledge decisions can be made on their own.
In this work Temporal Difference Learning is in the field of Reinforcement
Learning algorithms presented. For this, the most important practices are addressed and concrete examples are given. Its application is demonstrated by a video game, a variant of Asteroids. Furthermore, this implementation is analyzed and discussed extensively."

In this game, a bot (computer players) learns independently what it has to do.
It can choose from a range of actions like turning, flying and shooting.
If it shoots asteroids, it gets points. When it is hit, it gets points deducted.
These points will be added as a weighting to its random selection of actions,
depending on its environment (eg angle and distance to asteroids).
This slowly increases the probability that it's doing exactly that
what will deliver more points.


for Windows
Komprimiertes Archiv im ZIP Format 2.1 MB

Profil von Michael Grönert auf LinkedIn anzeigen


Michael Grönert


Michael Grönerts DevBlog anzeigen


Michael Grönert auf GitHub anzeigen


Profil von Michael Grönert auf DeviantArt anzeigen