DeMaMech: EU-Japan Collaboration in Higher Education Student Exchange Program

Offer of Research Topic

Offer from	University	Osaka University
	Faculty/Department/Institute	Department of Adaptive Machine Systems, Graduate School of Engineering
	Group Name	Architectonic Cognitive Science
Research Theme		Simultaneous Cooperative/Competitive Behavior Acquisition in Multi-Agent System
Brief Explanation		RoboCup has been proposed as a test-bed for multi robot systems in which a team of robots play a soccer game to beat an opponent team and most difficult task for the multi robot systems due to its highly dynamic, hostile situations. In such an environment, the robots have to adapt their behaviors to situations rapidly, or lose the game. In this study, a variety of learning/adaptation algorithms will be applied to a robot system participating in RoboCup MSL.
Research Steps		1. Literature survey (one month) 2. learning and improvement of simulator (one month) 3. Implementing various learning/adaptation algorithms on the simulator (two months) 4. Analysis of results (one month)
Period		5 months, but can be extended as a 10-month project, by including hardware experiments.
Necessary Background, Knowledge, Skills		Fundamental knowledge about robot systems and machine learning. Computer programming skills. Interest in RoboCup (www.robocup.org). If necessary, some references can be specified.
Further Information		http://www.er.ams.eng.osaka-u.ac.jp/
Contact Person		Dr. Yasutake Takahashi (yasutake@er.ams.eng.osaka-u.ac.jp)