隨著電腦與影像處理技術的逐步發展，影像視覺被廣泛應用在工業自動化中，對於檢測、定位與監督等工作都有其重要性。機械手臂為工業中常見的自動化設備，透過與影像視覺結合，建立機械手臂視覺系統，其具備活動範圍廣、可塑性高，並能因應不同的環境狀況等優勢，若是搭配其他設備工具更可增加其便利性與效率。 本研究乃是利用機械手臂與單攝影機建立3D立體影像(Stereo mage)， 針對曲面工件上螺孔進行姿態量測(位置與方向)，進而完成機械手臂之定位控制。本論文使用非接觸式立體視覺法，不須加設雷射光源，將兩張不同視角影像進行特徵點匹配找出現實空間中一點分別投影至二維影像的投影點，藉由計算兩張二維影像的投影點之間的視差，並利用現實空間與攝影機坐標系彼此間的幾何轉換還原此點的三維深度資訊。實驗驗證上，使用MITSUBISHI RV-2SD的六軸機械手臂及一台CCD攝影機以建立立體視覺系統，並以Visual Studio C++為基礎建立人機介面，將其實現於控制機械手臂自動鎖附螺絲於曲面工件上之螺孔。 ;With the development of computer and image processing technology, image vision is widely used in industrial automation. It is important for testing, positioning and supervision. The robotic arm is a common automation equipment in the industry. It combines with the image to establish a machine vision system. It has a wide range of activities, high plasticity, and can adapt to different environmental conditions. This research used a single camera and robotic arm to create 3D stereoscopic images, and the position and direction of the screw hole were measured to complete the positioning control of the robot arm. This research used non-contact stereo vision without the need for a laser source. The three-dimensional model of the screw hole is established by the principle of stereo vision to restore the lost depth information. In the experimental verification, the MITSUBISHI RV-2SD, six-axis robotic manipulator and a CCD camera were used to realize the stereo vision system. The Human Machine Interface is designed in C++ to control the manipulator lock screw and automatically attach the screw to the screw hole in the curved surface.
