I am currently developing automated players that use machine learning techniques to play my 3D soccer game. This is my final year project for a computer science degree at the University of Edinburgh, Scotland. I have implemented two types of automated players for the game. One type uses an explicit strategy that I have designed. The other type uses neural networks to learn basic behaviors and reinforcement learning to develop a higher level strategy. For anyone that is interested, my project report will be available for download here when it is done in a month.
If you have 15 minutes to participate in my survey, it would assist the further development of the reinforcement learning agents. Please download the program below and complete the brief online survey. The survey involves evaluating two different types of agents, one of which uses machine learning, on various criteria.
Note: The machine learning agent will be "learning" as you play, but you may need to play for several hours to notice it adapting to your strategy!
Download (Windows DirectX Version)
Complete The AI Agent Survey (after playing game and selecting survey option)
If you get a message that d3d9.dll was not found, please install DirectX 9.0 or higher.System Requirements:
Older Projects (Invasion of the Killer Kangaroo Balls)