The cooperation of multi-agent systems in an adversarial environment is an important research area of mobile robotics. There are several forums for researchers of the area to redound the international cooperation and the exchange of knowledge, such as the RoboCup competitions. Playing robot soccer is a complex scientific problem making it an attractive test-bench for the multi-agent artificial intelligence algorithms in real environment.
The subject of this thesis is the hardware and software system of a robot soccer player, which provide a suitable platform for further research on mobile robotic navigation and high level artificial intelligence. The application domain is the RoboCup Small Size League (SSL), the designed system complies with the 2013 SSL rules. In this document the created architecture is discussed, including the description of all the implemented components.
The real-time motion planning of multiple robots on a field with other rapidly moving obstacles is a complex problem. Several path planning algorithms exist for static environments, however they become less effective or computationally too expensive when applied with quickly moving obstacles. Two previously published methods for a dynamic environment are discussed and compared in this thesis. Both of the algorithms were implemented and tested in a simulator for the RoboCup SSL domain.