本論文針對2010年世界盃機器人足球競賽人形機器人足球比賽之規則提出了一套立體視覺自定位系統(Stereo Vision-Based Self-Localization System, SVBSLS)，此足球機器人將整合頭部水平與垂直方向馬達和立體視覺資訊，完成機器人於足球場上之自定位與量測足球之距離。此系統演算法開始先利用三角函數公式找尋機器人與地標、足球粗略的距離後，更進一步的採用類神經網路技術以提升距離的精準度，接著利用統計方法，求得人形機器人相對於地標的方位。 本研究著重在如何建構一套符合各種型號的攝影機且有效的機器人定位系統。一般以影像方式所完成之定位系統往往需要攝影機的內部參數，在此，我們利用一套有系統的方法自行量測攝影機之內部參數。根據上述所言，我們將能使用任何種類的攝影機並且準確的估測機器人在足球場上的座標與足球之距離。 實驗結果顯示，本論文所提出之SVBSLS定位系統，人形足球機器人自定位之精準度達到100%；人形足球機器人與足球距離平均誤差為0.64公分。 This thesis proposes a new Stereo Vision-Based Self-Localization System (SVBSLS) for the RoboCup soccer humanoid league rules of the 2010 competition. The humanoid robot integrates the information from the pan/tilt motors and stereo vision to accomplish the self-localization and measure the distance the soccer. The proposed approach uses the trigonometric function to find the coarse distances from the robot to the landmark and the robot to the soccer, and then it further adopts the artificial neural network technique to increase the precision of the distance. Next the statistics approach is used to calculate the relationship between the humanoid robot and position of the landmark. This research also focuses on how to establish an efficient robot localization system by using any type of CCD cameras. Generally the image-based localization system needs the intrinsic parameters. For the CCD camera adjustment, we propose a systematic method to measure the intrinsic parameters. By this approach, any type of CCD camera can be used to calculate the distance from the coordinates of robot on soccer field to the soccer. The experimental results indicate that the localization system of SVBSLS in this work has 100% average accuracy ratio for localization. The average error of distance from the humanoid soccer robot to the soccer is only to 0.64 cm.
