It's six years since I did my university final year project, but I'm still quite proud of it. I've meant to get around to publishing it online for a while. There is a copy in Leeds University Library, but I thought this would improve my readership.

I might put up my source files at some point, but for now I'm just including the report.

I developed a genetic computer algorithm that mimicked evolution to evolve small virtual robots that chased each other.

Summary

This report details the methods and results of a set of experiments into evolving virtually intelligent agents capable of displaying pursuit evasion behaviour. The advantages of randomness in such behaviour was investigated and how the complexity of the world affected agent behaviour was also looked into. The conclusions drawn were found to be similar to the ones expected but some interesting questions were also raised by the results.

Download the report

Evolving Pursuit and Evasion Strategies in Realistic Worlds ( 908KB, 42 pages)

I also used this project report as an excuse to try out testing PDFs for accessibility. It took me several hours to make this PDF accessible. This was mainly because there are quite a few tables in it. I've only been able to test it using the Adobe tools, so if anyone using a screen reader wants to try and let me know how they did, that'd be great.