- 學位論文
關節式機械手臂結合機械視覺之自動夾取系統研發
Development of Articulated Robotic Arm Integrated with Machine Vision and Automated Gripping System
摘要
隨著產業自動化趨勢的興起，機械手臂結合機械視覺技術逐漸取代人力，在生產線上扮演重要的角色。本論文旨在開發一套整合六軸機械手臂、RGB-D深度攝影機以及影像處理技術的系統，實現對隨機放置之特定物件的自動辨識與夾取。透過深度攝影機獲取RGB影像與深度影像，再使用Emgu CV函式庫將RGB影像進行HSV色彩空間轉換、二值化、膨脹等影像前處理，後續以多邊形擬合進行輪廓檢測，有效地辨識到特定物件的特徵與其三維位置。再將特定物件的於影像中的位置轉換成機械手臂的座標位置，最後經上位控制器控制機械手臂之氣動軟性夾爪實現對特定物件的夾取。本研究以眼在手的整合架構，為機械手臂提供了視覺感知的能力，成功實現了隨機放置物件之自動夾取系統，期望能應用在工業或農業中進行自動取放料之工作。
並列摘要
With the rise of industrial automation trends, robotic arms integrated with machine vision technology are gradually replacing human labor, playing a significant role on production lines. This thesis aims to develop a system that integrates a six-axis robotic arm, an RGB-D depth camera, and image processing technology to achieve automatic recognition and picking of randomly placed specific objects. The system obtains RGB and depth images through the depth camera, then processes the RGB images using the Emgu CV library for HSV color space conversion, binarization, dilation, and other preprocessing techniques. Subsequent polygon fitting for contour detection effectively identifies the features and three-dimensional positions of specific objects. The position of these objects in the images is then converted into the coordinates for the robotic arm. Finally, a pneumatic soft gripper controlled by an upper-level controller implements the picking of specific objects. This research adopts an "eye-in-hand" integrated framework, providing the robotic arm with visual perception capabilities, successfully realizing an automated picking system for randomly placed objects, with potential applications in industrial or agricultural automated material handling.