簡易檢索 / 詳目顯示
| 研究生： |
姜育欽 Jiang, Yu-Chin |
|---|---|
| 論文名稱： |
六自由度機械手之設計、實作與視覺伺服應用於按鈕按壓控制 Design, Implementation and Visual Servo of a 6-DOF Robotic Arm for Button Pressing Control |
| 指導教授： |
葉廷仁
Yeh, Ting-Jen |
| 口試委員: |
林沛群
Lin, Pei-Chun 劉承賢 Liu, Cheng-Hsien |
| 學位類別： |
碩士 Master |
| 系所名稱： |
工學院 - 動力機械工程學系 Department of Power Mechanical Engineering |
| 論文出版年： | 2017 |
| 畢業學年度： | 105 |
| 語文別： | 中文 |
| 論文頁數： | 56 |
| 中文關鍵詞： | 影像辨識 、立體視覺 、機械手臂 、伺服控制 |
| 外文關鍵詞： | image recognition, stereo vision, robot arm, servo control |
| 相關次數： | 點閱：2 下載：0 |
| 分享至: |
| 查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
本研究主要在設計與發展一套能經由影像伺服控制之六自由度機械手，進而達成如按壓按鍵任務的系統。本系統於手臂基座安裝一個固定的雙眼攝影機，並由此攝影機擷取影像資訊，使得機器人得以辨識環境中的目標物，以及獲取目標物在三維空間的狀態。在影像辨識部份，將採用加速穩健特徵點演算法(Speed up robust feature, SURF)來取得目標物的特徵點，為了優化特徵點比對結果，本系統亦採用隨機抽樣一致演算法(RANdom SAmple Consensus, RANSAC)來除去錯誤比對的離群值，以計算出較為準確的單應性矩陣(Homography matrix)，藉此獲取目標物在當前影像中的位置。在立體視覺部份，我們可以藉由影像辨識在左右兩幅影像中所找到的對應點，經由三角幾何的關係還原目標物的深度資訊。在機械手部份，我們將使用3D列印及部份零件自製六軸的機械手臂，並自行設計控制器對關節進行伺服控制，機械手臂末端位置將使用閉路雅可比轉置方法求解逆向運動學。在指尖末端也會放置力敏電阻以感測按壓按鈕力道。最終將使機器人藉由上述方法完成執行按壓按鈕的任務。
This study is mainly in the design and development of a 6-DOF robotic arm for the task such as pressing a button on an elevator panel. There is a binocular camera fixed near the base of our robotic arm. Thus, the robot can obtain the image information from the camera, such that it can recognize a target in the scene, and can calculate its pose in 3-D space. In the section of image recognition, we adopt an algorithm called “Speed Up Robust Feature” to obtain the features. In order to optimize the matching results, we use “RANdom SAmple Consensus” to remove the outliers caused by the wrong matching results and get a more correct homography matrix to obtain the position of the target in the current image. In the section of stereo vision, corresponding points can be identified and found in the left and right images by image recognition. The depth information of the target can also be reconstructed by the triangular geometry. In the section of robot arm, there are some components made with 3-D printer. The servo control of joints also adopt a self-designed controller, and we use closed-loop Jacobian transpose scheme as the inverse kinematics algorithm. A force-sensitive resistor is also placed at the end of the fingertip to measure the force caused by pressing a button. Eventually, the robot can use techniques above to complete the task of button pressing.
[1] 林玗瑾. "輪型機器人之影像辨識應用於跟隨人與電梯辨識."中央大學電機工程學系學位論文(2012): 1-77.
[2] 張哲豪, 宋開泰, and 林清安.基於立體視覺之移動式機械臂影像伺服設計. Diss. 2009
[3] Lowe, David G. "Distinctive image features from scale-invariant keypoints." International journal of computer vision 60.2 (2004): 91-110.
[4] Bay, Herbert, et al. "Speeded-up robust features (SURF)." Computer vision and image understanding 110.3 (2008): 346-359.
[5] Juan, Luo, and Oubong Gwun. "A comparison of sift, pca-sift and surf." International Journal of Image Processing (IJIP) 3.4 (2009): 143-152.
[6] Fischler, Martin A., and Robert C. Bolles. "Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography." Communications of the ACM 24.6 (1981): 381-395.
[7] Zhang, Zhengyou. "A flexible new technique for camera calibration." IEEE Transactions on pattern analysis and machine intelligence 22.11 (2000): 1330-1334.
[8] MaykelMoya ,“camera_calibration,”2014. [Online]. Available: http://wiki.ros.org/camera_calibration
[9] Chiacchio, Pasquale, and Bruno Siciliano. "A closed‐loop jacobian transpose scheme for solving the inverse kinematics of nonredundant and redundant wrists." Journal of Field Robotics 6.5 (1989): 601-630.
[10] “Aldebaran documentation”[Online]. Available: http://doc.aldebaran.com
[11] “FSR 402 Data Sheet”[Online]. Available: http://www.trossenrobotics.com/productdocs/2010-10-26-DataSheet-FSR402-Layout2.pdf