[[alternative]]Billiard Robot(III)

計畫編號：NSC98-2221-E032-036研究期間：200908~201007研究經費：636,000[[abstract]]本研究重點是整合目前既有的研究成果，其中包括以灰色理論、模糊理論、可拓理論和倒傳遞類神經演算法發展出的各種撞球球局攻擊及防守決策。令撞球機器人能分別針對需要採取攻擊或防守(安全球)的球局以多種技巧和不同決策進行擊球進袋或是解球的動作，並應用人工智慧進行各項決策後，搭配氣壓缸及氣壓控制閥等硬體設備實際呈現撞球機器人的進攻、解球及防守等全方位功能。 首先利用數位影像處理技巧，改善影像模糊的問題以及對於各號子球的辨識，使得撞球機器人能夠依照花式撞球中的九號球規則，再由幾何關係依照子球、母球及球袋的相關位置判斷出擊球角度，選擇最易進球的球袋。再以模糊理論補償擊球角誤差與控制擊球力道，以提昇撞球進球率，依序將目標球撞進球袋，達到贏球的目的。 如果無法順利攻擊，則必須考慮解球或防守，針對九號球障礙球局，為使撞球機器人具備撞球選手的解球思維與推理能力，利用可拓理論建立決策機制，分析出最佳解球模式。並從球檯邊與球的幾何關係，算出顆星撞擊點，再以所設計的VB程式，預測可能的撞擊點及撞擊後母球的停留位置，為避免在球局中犯規，另外加入母球/子球是否撞擊檯邊的判斷，來偵測最佳的撞擊點，以完成解球動作。 另外以灰色理論建立一套智慧型撞球機器人之防守策略機制，加入障礙球之偵測，針對具有障礙球的情況以不同之防守策略來增加對手進球難度，然後決定最不利於對手進球之預測位置，再以此位置反向推論撞球機器人之打擊方式，驅動撞球機器人打擊，完成防守動作。以提高撞球機器人於球局判斷之能力，即撞球術語中之「安全球」，使撞球機器人能反「守」為「攻」，因而贏得球局。 The objective part of the research is to integrate the research results of past few years, and to develop different offensive and defensive strategies in billiard game by intelligent algorithms ( ex: gray theorem , fuzzy logic theorem , extension theorem or neural network algorithm). For the different purposes of billiard games, we can let the robot to pocket object balls into corresponding pockets, or make a safety play by the decision-making subsystem. The shooting command is executed by a pneumatic cylinder and a control servo-volve when the motion decision is done. Billiard Robot can make offence shot, defense shot and position play to achieve different purpose practically. First, we can get image data form CCD camera, and to recognize the balls from the background by building color model of RGB values and pixel discrimination. We can establish the AI system by VB to simulate how people play billiard. The AI system is applied to decide the strength and angle of shooting from the ball positions and to improve the skill and to increase the rate of pocketing with the fuzzy theory that helps the error compensation and strength control. If we can not make the offence shot, a safety play is considered by this robot. The objective of this section is to develop a strategy by Extension theory to make billiard robot posses the imitation ability of how people let the cue ball to contact the objective ball after shooting in the block ball game of the nine-balls pool games. In addition, we develop a defensive strategy of the intelligent billiard robot by using the grey theory. The main purpose of this defensive strategy is to make a safety play during the pool game. The information of distance between the object ball, cue ball, corresponding pocket and the block ball are calculated by this defensive strategy. Then, the billiard robot will execute the hitting command to make a safety play based on this algorithm. The safety play is the defensive positioning of the balls so as to minimize the opponent’s chance to score or to make a foul stroke. In this case, it will be easier for us to win the ball game whenever opponents have to face to a safety play.[[sponsorship]]行政院國家科學委員