Seminar: A Hybrid Genetic Programming Based Decision Making System for Multi-Agent Systems of RoboCu

Amir Tavafi
M.Sc. Thesis Proposal
Supervisor: Dr. Wolfgang Banzhaf


A Hybrid Genetic Programming Based Decision Making System for Multi-Agent Systems of RoboCup Soccer Simulation


Department of Computer Science
Thursday, January 28, 2016, 1:20 p.m., Room EN 2022


Abstract

In this proposal, a hybrid genetic programming based approach is proposed for decision making system in the complex multi-agent domain of RoboCup Soccer Simulation. In the past, genetic programming was rarely used to evolve agents in this domain due to the difficulties and restrictions of the soccer simulation domain. The proposed approach consists oftwo phases each of which trying to cover the other one’s restrictions and limitations. The first phase will produce some evolved individuals based on an offline GP algorithm and the second phase uses best individuals of first phase as an input to run a GP algorithm in order to evolve players in the real game environment where evaluations are done during real runs of the soccer simulator. It is expected that the evolved individuals after the second phase are able to outperform a team of the same quality skills with a static decision making system. All of the implementations and experiments will be done using source code of the award winning and internationally recognized team, “MarliK” that was developed and lead by the researcher of this thesis.

 

Contact

Department of Computer Science

230 Elizabeth Ave

St. John's, NL A1B 3X9 CANADA

Tel: (709) 864-2530

Fax: (709) 864-2552

becomestudent@mun.ca