Ball Trapping in the Simulated Soccer using Decision Tree based Learning to Rank

In a group ball game such as soccer, the ball passing behavior between players is important for achieving cooperative team behavior. To acquire the ball passing behavior, conventional approaches mainly apply search and machine learning to the decision making of the players who perform the passing action. On the other hand, the position and posture of the pass receiver player when receiving the ball have not been studied sufficiently. This paper proposes a machine learning method using decision tree based learning to rank to select a more advantageous ball trapping behavior. We use the RoboCup Soccer Simulator as an experimental environment to collect training datasets and to evaluate the performance of the action selection model.