A New Passing Strategy Based on Q-Learning Algorithm in RoboCup

Abstract

The RobuCup 2D soccer simulation has been used as the basis for successful international competitions and research challenges and to simulate the interest of the public for robotics and artificial intelligence (AI). In cooperation strategy designing, researchers often employ the method of machine learning to optimize the simulation system, as the research growing, Q-learning algorithm, which is a particular type of machine learning, is becoming more popular. The paper is extended as follows: First, we will make a description of the characteristics about the RoboCup simulation system. Second, there is an analysis of the limitation about the traditonal Q-Learning algorithm, and a new strategy based on Q-Learning will be advanced here. Finally, Simulated results will be discussed and the paper comes to a conclusion.