本論文主要目的是為使撞球機器人具有類似撞球選手在球局中的解球思維與推理能力，除了整合直線打擊的能力於非障礙球局中進行攻擊之外，更以模糊理論建立撞球機器人在九號球局障礙球局中的顆星攻守決策系統，於障礙球局中判斷出攻擊或防守的最佳擊球選擇，在決定攻擊的情況下完成擊球進袋得分或是搭配可拓防守決策解球達成防守的目的。 撞球機器人先由CCD攝影機擷取球檯上的影像，搜尋各球的球心座標，判斷無障礙球之後採取直線進攻，而在有障礙球的情況下運用球檯邊與各球的幾何關係，計算顆星撞擊點，進而求得子母球距和攻擊角，並以擊球角、子母球距，作為模糊攻守決策的衡量條件，判斷應採取攻擊或是防守行動。採取攻擊則搭配顆星攻擊擊球進袋，而採取防守則以VB程式，預測可能的顆星撞擊點及力道撞擊目標球後的相關位置，以母球與虛擬撞擊點的距離、目標球與球袋的距離、擊球角、障礙球位置判斷等四個決策因子，作為可拓防守決策的衡量條件，使撞球機器人達到解球防守的目的，經模擬及實驗結果得知，顆星攻守決策系統確實能達成球局進攻或防守的目標，而直線攻擊、顆星攻擊及顆星防守三項打擊能力皆能達到70%以上的成功率。 The objective of this thesis is to develop a offensive/defensive decision-making system for the billiard robot by using Fuzzy theory. The main purpose is to make billiard robot possess the imitation ability of how human beings to do the offensive/defensive decision-making in the block ball game of the nine-balls pool games. First, the CCD camera captures the image of the balls on the pool table. Then, the positions of the balls are calculated by the image processing technique and the developed software program (VB). If there is no block ball between the cue ball and the object ball, the billiard robot will do the straight shooting to pocket the object ball. If there is a block ball between the cue ball and the object ball, the robot will analyze the hitting positions of cushion which are determined by the distances between the object ball, the cue ball and the rail of the table for decision-making. The system is developed by the following two conditions which are (1)the distance from the cue ball to the hitting positions of cushion and to the object ball, and (2)the offensive angle by the Fuzzy theory. The offensive strategy of cushion means to pocket the object ball. The defense strategy is developed by the following four parameters which are (1)the distance between the cue ball and the object ball, (2)distance between the object ball and the corresponding pocket, (3)the angles between the cue ball, the object ball and the corresponding pocket and (4)the information of the block ball by the Extension theory. In addition to the four parameters, it must be considered whether the cue ball or the object ball contacts the rail of the table after the cue ball contacts the object ball. Finally, the billiard robot will execute the hitting command to let the cue ball contact the object ball and make a safety play. The simulated and experimental results show that this developed offensive/defensive decision making system work very well in the pool game. According to the result of experiment, the successful possibility of shooting is more than 70%.
