'An AI Framework for Real-Time Strategy Games'

Studenteropgave: Kandidatspeciale og HD afgangsprojekt

  • Kasper Frederiksen
  • Kasper G. Kristensen
  • Anders Lauritsen
4. semester, Datalogi, Kandidat (Kandidatuddannelse)
'Recently computer gamers have started to focus more and more on the gameplay aspect of computer games. This has led to an increased interest in the multiplayer aspect of games as they present a challenge that the AI in the single player part cannot. So far the single player AIs have been easy to beat, because they lack dynamic capabilities. This project focuses on defining an AI framework architecture for the specific genre of Real-Time Strategy games, which will help to develop AIs with more human-like capabilities and behaviour in this genre. We present a framework based on a model of how a player plays Real-Time Strategy games. This framework is then tested by connecting it to a Game Development Framework. We show that the framework can be configured by both programmers and non-programmers, and that the framework can be used to provide complete AI solutions within the main stream RTS games. Preliminary results show that the ideas behind the framework design and a new data structure, introduced in the project for handling strategies, are very promising.'
ID: 61067782