Coaching: Learning and Using Environment and Agent Models for Advice

Ph.D. Thesis: Patrick Riley

This page serves as the online presence of my Ph.D. thesis.

Thesis Committee

• Manuela Veloso, chair
• Tom Mitchell
• Jack Mostow
• Milind Tambe, University of Southern California

Documents

• Abstract
• The whole document in pdf and gzipped postscript
• The CMU SCS tech report of my thesis. (coming soon)
• My personal publication description of my thesis
• The slides from my thesis defense
• If you are interested in hiring me, please see my resume or my CV.

Source Code

• Wyverns soccer team
• OWL Coach
• RCSSMaze environment (includes both the environment code and the agents I created)
• Predator-Prey environment and the shared coach code to go with it. This includes both the environment and agent code.
• Game Runner Maintains and runs a queue of RoboCup simulation games. I used variants of this system to run most of the games from the thesis.

Datafiles

• Experimental work was very important for this thesis, and I have about 94G of data files (mostly Soccer Server logs). If you would like this data, please send me an email. I should be able to arrange to send you some DVDs of data, but I may ask you to pay for media and shipping.

Misc. Links

• Carnegie Mellon simulated robot soccer
• Carnegie Mellon CORAL research group
• Back to my homepage