撞球機器人在九號球局之應用

Application of the Nine-Balls Pool Games for a Billiard Robot

黎哲廷(Je-Ting Li)

指導教授 ： 楊智旭

淡江大學/工學院/機械與機電工程學系碩士班/碩士(2005年)

https://doi.org/10.6846/TKU.2005.00025

摘要

本研究之目的是使用數位影像處理技術，對球檯上各號球進行影像辨識。使得撞球機器人能夠依照花式撞球中的九號球規則，依球號順序將號球撞進球袋。
在機器視覺方面，使用CCD攝影機擷取球檯上的影像，利用數位影像處理技術得到母球與各號球的確切位置座標，再配合決策系統來完成擊球順序規劃。在機器思維方面，以VB設計的判斷程式讓撞球機器人模仿人類學習撞球的思維，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。
撞球機器人應用可拓理論來模仿撞球選手的思考推理能力，其決策系統分為可拓球袋選擇決策與可拓作球決策，首先利用球袋選擇決策來避開障礙球，選擇最易進球的球袋，再利用可拓作球決策來預測母球在撞擊子球後的行進路線及可能的停留位置，路線上有障礙球也能預測母球與障礙球碰撞後的行進路線，來找出最容易撞進下一顆子球時母球的停留位置，並持續模擬撞進球檯上所有子球的進攻路線，做出全局進攻流程規劃，完成清檯(Clean table)的目標。經由實驗驗證，撞球機器人已經可以辨識球檯上的母球與1∼9號球，依照實驗設計，將1~4號球依序打擊，將子球撞進球袋。

關鍵字

撞球機器人 ； 數位影像處理 ； 機器視覺 ； 可拓理論

並列摘要

The objective of this thesis is to develop an offensive strategy of the nine ball game to score ball from number 1 to 9 sequentially for the billiard robot. The strength of shot is determined where the cue ball will stop in its trajectory to be in a good position for the next shot by the controller. The image pattern recognition technique is applied to identify the cue ball and each numbers of the object balls on the table. Then, the positions of the balls will be determined automatically by the machine vision subsystem and the developed software(VB). After that, the extension theory is applied to develop the offensive strategy of the nine ball game in order to win the game successfully.
There are five parameters which are needed to be considered together for the offensive strategy. They are (1)distances between the cue ball and each object balls, (2)distances between object balls and the corresponding pockets,(3)block balls between the cue ball and object balls, (4)the angles between the cue ball, object balls and the corresponding pockets, (5)the predicted positions of the cue ball and object balls after collision. Then the offensive strategy is developed based on the above five parameters by the Extension theory. Finally, the billiard robot will execute the hitting command(hitting position and strength of shot) to pocket the object balls into corresponding pockets one by one in order to win the billiard game.
The experimental results shows that the billiard robot is able to “read” the balls on the table correctly and pockets the object balls from the lowest number to the highest number one by one successfully.

並列關鍵字

Billiard robot ； Digital image processing ； Machine vision ； Extension theory

參考文獻

[3] T. Jebara, C. Eyster, “Augmenting the Billiards Experience with Probabilistic Vision and Wearable Computers,” IEEE Wearable Computers, Oct 1997 ,Vol.13 pp. 138 – 145.

[6] H. Nakama, I. Takaesu and H. Tokashiki, “Basic Study on Development of Shooting Mechanism for Billiard Robot”, JSME, Robotic Workshop, 2001, 1A1F8.

[16] 盧彥宏，撞球機器人在撞球球局之防守決策，淡江大學機械與機電工程學系碩士論文，民國94年6月

參考文獻

[1] 張貴敏、陸仲元、張萍、張秀萍編著，撞球，體育運動叢書，1999年8月。