Behavior Based Fuzzy Logic for World of Warcraft

Studenteropgave: Speciale (inkl. HD afgangsprojekt)

cover.pdf
  • Jon List Thorsteinsson
  • Søren Møller Larsen
4. semester, Software, Kandidat (Kandidatuddannelse)
This project covers the development of an autonomous agent capable of playing against human players in World of Warcraft, and is an extension of work done in the previous project.
An expert system that combines a behavior based approach with fuzzy logic has been developed, in which the fuzzy rules are weighted. The weights of the rules are adjusted with reinforcement learning to ease the configuration.
The experimental results shows that it is easy to configure an agent in the expert system and that the weights of the fuzzy rules are learned within a short period of time. The agent is far superior to the previously developed scripted agent and is capable of challenging experienced human players in a 3 versus 3 player scenario.
SprogEngelsk
Udgivelsesdato8 jun. 2012
Antal sider69
ID: 63661651