ปัญญาประดิษฐ์สาหรับการส่งบอลโด่งในเกมฟุตบอลแบบสามมิติ / ณัฐวิชช์ ตัญจพัฒน์กุล = Artificial intelligence for long ball passing in 3d football game / Nattawit Tanjapatkul
Ball passing is very important in soccer game. Making decision to pass the ball to the right place and right time can help increasing the scoring chance for team. There are many previous researches in RoboCup which are based on ball passing. But those researches were using 2D soccer simulation. This limited AI to only pass the ball on the ground aka normal passing. In real soccer game, long ball passing also has an important role. If AI can pass the ball in both styles, it will increase the performance and will have many varieties to play soccer. Although, soccer games in today market are very realistic and AI is capable of both normal passing and long ball passing. But the development technique behind it is not proposed to the public. So, the main objective of this thesis is to propose the method to develop ball passing decision, especially long ball passing, in 3D environment for AI. We propose the algorithm to calculate suitable kicking force for ball passing. We have developed our own 3D simulation soccer using Unity3D game engine. We apply Pareto Optimality in ball passing decision in order to search for suitable passing destination for normal passing and long ball passing. Furthermore, we also propose the method to choose each passing type by applying dynamic script concept. Because, even if, AI can pass the ball in various types. But to achieve the best performance, AI must decide which passing type is more appropriate for certain situation. The results show that AI using our proposed method has better performance than AI in previous research. We believe that our proposed method can be used in real soccer game and can also be applied to other sport games which have team collaboration like soccer.
