Robotsoccer team development in RoboCup Environment

OData support
Dr. Harmati István
Department of Control Engineering and Information Technology

The development of robot soccer agents is an important field in robotics, since it requires applying a broad range of technologies from solving control theory problems to image processing techniques. It is a convenient platform for studying the nature of cooperative systems in a dynamic and competitive environment. However, building a whole team of robots is expensive and the researchers have to allocate a lot of resources for problems on the phisycal level, instead of focusing on the algorithms of strategy and cooperation.

The international ground for these studies is the software simulator platform and the annual championship of the RoboCup Initiative. The platform helps researchers by automatically solving the low-level control and image processing problems and simulating the appropriate environment for the software agents. As a result, clients built for the platform can be used to develop, analyze and test high level strategies and cooperation.

Of course agents that can play in the simulation have to cope with lots of problems such as communicating with the server, navigating autonomously or calculating position and orientation. These are non-trivial tasks and require the use of different software technologies.

The ultimate goal is building a team of autonomous agents that can perform actions to reach long term goals. To accomplish this, on the high level strategic and tactical level a number of problems have to be solved, such as trajectory planning or fuzzy control. The MATLAB system and programming language is an efficient and widely used tool for solving these kind of problems. However, it is a non-trivial task to write an autonomous agent in this language, that can play obeying the rules of the platform and I know of no such software client up to date that can exploit the advantages of MATLAB.

In my thesis I propose a software client developed for this platform, which enables one to test strategic and tactical layers of multi-agent cooperative systems developed in the MATLAB language. I also propose a tactical layer implemented in MATLAB and tested using the client software.


Please sign in to download the files of this thesis.