Adaptive Reinforcement Learning Multiagent Coordination in Real-Time Computer Games

From AIRWiki
Jump to: navigation, search

Part 1: project profile

Project name

Reinforcement Learning Multiagent Coordination in Adaptive Real-Time Computer Games

Project short description

This aim of the project is to obtain adaptive behaviours in real-time videogames. We focused on multiagent "enemy" coordination, using a distributed optimization algorithm designed from the COIN framework. The project is based on an original game developed using Microsoft XNA, a free C# framework created to develop games on both PC and XBOX360. The reinforcement learning part is based upon a Windows port of PRLT

Dates

Start date: 2007/09/01

End date: 2008/07/22

Internet site(s)

http://www.bloodymonkey.com/

http://prlt.elet.polimi.it/mediawiki/index.php/Learning_in_Computer_Games

http://www.xna.com/

http://creators.xna.com/

People involved

Project head(s)

Marcello Restelli - restelli (at) elet (dot) polimi (dot) it

Alessandro Lazaric - lazaric (at) elet (dot) polimi (dot) it

Other Politecnico di Milano people

Students

Students currently working on the project

Marco Canala - marco (dot) canala (at) gmail (dot) com

Paolo Tajè - paolo (dot) taje (at) gmail (dot) com

Students working on the project in the past

Laboratory work and risk analysis

Laboratory work for this project will be mainly performed at AIRLab. The only risky activity is the (necessary) use of Windows Xp and C#/.NET.

Part 2: project description