English  |  正體中文  |  简体中文  |  Items with full text/Total items : 49962/85138 (59%)
Visitors : 7792336      Online Users : 54
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library & TKU Library IR team.
Scope Tips:
  • please add "double quotation mark" for query phrases to get precise results
  • please goto advance search for comprehansive author search
  • Adv. Search
    HomeLoginUploadHelpAboutAdminister Goto mobile version
    Please use this identifier to cite or link to this item: http://tkuir.lib.tku.edu.tw:8080/dspace/handle/987654321/103094


    Title: 視覺伺服控制機械手臂系統,-子計畫四:以CAD模型為基礎之視覺追蹤系統的設計與實現(I)
    Other Titles: Design and Implementation of a Cad Model-Based VI Sual Tracking System
    Authors: 蔡奇謚
    Contributors: 淡江大學電機工程學系
    Keywords: 工業機械手臂;視覺追蹤;姿態估測;CAD 模型建置;CPU-GPU 異質運算;Industrial robot manipulator;visual tracking;pose estimation;CAD model building;CPU-GPU heterogeneous computing
    Date: 2012-08
    Issue Date: 2015-05-18 16:12:43 (UTC+8)
    Abstract: 由於工業機械手臂必須在一個複雜的工作環境中進行準確的自主式移動規劃及伺服控制,透過視 覺感測求得物體姿態資訊是重要的基礎功能之一。本計畫的主要目的為研發一個以三維姿態估測技術 為基礎之視覺追蹤系統,用來即時提供機械手臂於任務執行中所需要的物體種類與三維姿態資訊,藉 此提高機械手臂的移動規劃與伺服控制能力。為了達到此目的,除了穩定且快速的視訊增強技術之 外,影像處理中的影像物體分割、物體姿態估測、物體移動追蹤以及系統運算效能等問題也是必須要 考量的。在視訊增強技術方面,為了能提供穩定的影像品質,在影像前處理中必須加入即時色彩與區 域對比增強演算法。在影像物體分割方面,可採用以色彩資訊為基礎的影像分割演算法,提供即時的 影像物體偵測結果。在姿態估測演算法方面,為了能提供強健的估測結果,可採用以CAD 模型為基 礎之三維姿態估測演算法,不但能提高視覺追蹤系統的強健性,同時也能提供物體的完整三維姿態資 訊。而在CAD 模型建置方法上,為了避免人工3D 繪圖方式所造成之誤差的影響,系統必須搭配一個 可自動建置物體CAD 模型的演算法,用來自動建置三維視覺追蹤系統所需要的物體CAD 模型。在物 體移動追蹤方面,除了可透過姿態估測結果進行預測外,也可透過以影像為基礎之視覺追蹤法進行移 動預測任務,藉此提高追蹤效能。在提高系統運算效能部分,由於三維視覺追蹤系統的處理過程中將 需要使用大量的運算,若使用傳統的CPU 運算架構將會影響整體系統的即時性。因此,為了維持系統 的運算即時性並提高輸出的更新率,透過CPU 與GPU 整合運算的CPU-GPU 異質運算技術將會是重要 的關鍵技術。藉由以上技術的結合,本計畫將預期發展出一套以CAD 模型為基礎之視覺追蹤系統, 用來作為自主式機械手臂系統之視覺追蹤與物體姿態估測裝置。我們預計以兩年時間完成此計畫,第 一年之重點在影像前處理功能、物體初始姿態估測以及自動CAD 模型建置演算法的設計,第二年將 集中在物體姿態估測與物體移動追蹤功能的實現,並將系統整合於自主式機械手臂中,同時進行系統 測試與驗證。
    Since an industrial robot manipulator usually works in a complex working environment, the function of object pose estimation through visual sensing becomes one of the important fundamental functions in order for the robot to perform motion planning and servo control accurately. This project aims to develop a 3D pose estimation-based visual tracking system for industrial robot manipulators to handle object tracking and pose estimation problems during motion planning and servo control tasks in a complex working environment. To achieve this, the visual tracking system requires developing several key techniques such as video enhancement, image segmentation, object pose estimation, and moving object tracking, etc. For the design of video enhancement, a real-time color enhancement with local-contrast preservation algorithm is demanded to improve image quality. For the development of image segmentation algorithm, since color information is an efficient and robust feature to extract multiple objects simultaneously, it is preferred to develop a color-based object detection algorithm. In the design of pose estimation algorithm, a 3D model-based pose estimation algorithm is able to provide the 3D pose information of an object of interest and improve robustness of the visual tracking system. Such a system needs a CAD model building algorithm to generate an accurate CAD model of the object of interest. In order to improve the performance of moving object tracking, it is better to apply an image-based visual tracking algorithm to improve tracking performance of the system. Moreover, in order to achieve real-time computing performance, the CPU-GPU heterogeneous computing technology is helpful to improve the computation efficiency of overall 3D pose estimation algorithm. A model-based visual tracking system with real-time computing capability can be constructed by the combination of the techniques discussed above. Such system can be used as a 3D visual sensing system for an industrial robot manipulator working in a complex environment. We plan to complete this study in two years. The first year will focus on the image preprocessing, pose estimation algorithm and point cloud data processing. In the second, the emphasis will be on the implementation of CAD model building, object motion prediction, and object classification. The developed visual tracking system will be integrated into two robot manipulators to evaluate the performance in practical experiments.
    Appears in Collections:[電機工程學系暨研究所] 研究報告

    Files in This Item:

    There are no files associated with this item.

    All items in 機構典藏 are protected by copyright, with all rights reserved.


    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library & TKU Library IR teams. Copyright ©   - Feedback