Robot Soccer

Abstract

This doctoral thesis deals with the robot soccer domain and it is mainly focused on the area of strategy planning, analysis, optimization and adaptation. The main goal of the thesis is to create new algorithms and methods which will be applicable in different simulators and environments; not only in the area of robot soccer games. This work describes the methods for the strategy evaluation, loop strategy detection and hub rule detection. The methods utilizing graphs and space filling curves for the fast and effective rule selection from the strategy; and methods for sequence extraction and comparison to validate and visualize the strategy adaptation mechanism are presented. This thesis also describes the custom robot soccer architecture which was created to test the proposed methods and algorithms. The proposed methodology can be used in other thematic areas, where there is a need for an effective strategy planning.