Robot Soccer Using Deep Q Network

摘要

Reinforcement Learning is one of brilliant way to develop intelligent agents in the field of Artificial Intelligence. This paper proposes a RL algorithm called Deep Q Network and presents applications of this algorithm to the decision-making problems challenged in the RoboCup. Four scenarios were defined to develop decision-making for a SSL in various situations using the proposed algorithm. Furthermore, a Convolutional Neural Network model was used as a function approximator in each application. The experimental results showed that the proposed Reinforcement Learning algorithm had effectively trained the Reinforcement Learning agent to acquire good decision making. The Reinforcement Learning agent showed good performance under specified experimental conditions.