Recognizing Patterns of Dynamic Behaviors Based on Multiple Relations in Soccer Robotics Domain

Abstract

This work is focused on the recognition of team patterns represented by different formations played by a soccer team during a match. In the soccer domain, the recognition of formation patterns is difficult due to the dynamic and real time conditions of the environment as well as the multiple interactions among team mates. In this work, some of these multiple interactions are modeled as relations represented by a topological graph which is able to manage the dynamic changes of structures. Thus, the topological graph serves to recognize apparent changes of formations from real changes of them. The proposed model has been tested with different teams in different matches of the Robocup Simulation League. The results have shown that the model can recognize the different main formations used by a team during a match even the multiple changes of the players due to the dynamic nature of a match.