2021 – TDP – KIKS – RoboCupSoccer – Small Size

This paper is used to qualify as participation to the RoboCup 2020 soccer small size league. Our team’s robots and systems are designed under the RoboCup 2019 rules. In this paper for hardwear, it is described about the various idea of dribble mechanism, whose importance was again emphasized in the final of RoboCup 2019 Sydney. In software, the implementation of UCT (Upper Confidence bound applied to Trees) in MCTS (Monte Carlo Tree Search) method was performed to achieve the action-dicision of the motion for the robots. Moreover, we tried to use Deep Learning Library (Sony products) and Neural Network using GUI (Graphical User Interface) for layered design, learning and evaluation in AI system.