臺灣博碩士論文加值系統

在工業4.0的趨勢下，無人化工廠概念興起，政府也在推動智慧製造政策，期望藉由開發無人智慧工廠以因應目前缺工之情況，因此大多數工業領域產線無人化缺少不了機械手臂協助取放料，現今大多數加工產業運用機械手臂進行取料，需先將工件整列擺放或以2D視覺影像搭配輸送帶進行運輸，前置作業較為麻煩且耗時，靈活度較低，如此將限制機械手臂發展空間的可能性。故本研究開發一套機械手臂於堆疊物件之夾取系統，本系統可在堆疊物件中辨識出料件正反面姿態，並透過機械手臂夾取無遮擋且高度最高之最佳夾取料件。本研究論文主要分成三大部分，首先為結構光掃描系統，基於格雷編碼原理進行結構光圖形編碼，搭配相機與投影機進行編碼圖形的投放與擷取，利用線與線三角量測原理，將工件胚料外型進行點雲建模，並取得物件三維點雲圖形資訊；其二為估測堆疊物件姿態系統，透過結構光掃描系統建立物件三維點雲模型，藉由三維點雲處理技術，計算樣本點雲與堆疊物件點雲三維法向量、點對特徵描述匹配(Point Pair Feature, PPF)與ICP(Iterative Closest Point )匹配等三維點雲演算法，辨識出堆疊物件中目標物不同方向與姿態，求得出物件的6-DOF姿態變化量，並由實驗得出姿態估測系統精準度為線性軸誤差1.5mm與旋轉軸誤差3度以內；其三為機械手臂智慧夾取系統，透過估測堆疊物件姿態系統，估測出物件6-DOF姿態後，透過新代科技所提供之Remote API函式庫與新代手臂連線通訊，將估測出的物件六軸姿態上傳機械手臂控制器中，並撰寫新代PLC(Programmable Logic Controller)與MACRO程式，控制氣動夾爪I/O點開關與規劃機械手臂路徑，使機械手臂能夠辨識出堆疊雜亂中的物件，並且夾取料件至指定位置擺放，達到機械手臂智慧夾取之功能，提高機械手臂在工業產線上的價值與靈活性，達到智能化與無人化產線之目的。

Under this Industry 4.0 tendency, the concept of the unmanned factory has risen. The government is also promoting smart manufacturing policies which hope to respond to the current lack of workers by developing unmanned smart factories. In most industrial fields, the unmanned assembly line is indispensable for the robot arm to assist in picking. Nowadays, processing industries use robot arms for picking and placing. It is necessary to arrange the workpieces in a row or transport them with 2D images and conveyor belts. The front-end operation is troublesome time-consuming and low flexibility which will limit the development of the robot arm. Therefore, this study develops a system for robot random bin-picking of multiple objects. The system can identify the front and back pose of the object from the multiple objects and use the robot arm to grasp the best object. This study is mainly divided into three parts. Firstly, the structured light system is based on the gray coding principle which can use a camera to capture the coded patterns projected by the projector. The system uses the principle of the line-line intersection to model objects in three dimensions which obtain 3D point cloud information of the object. The second is the pose estimation system of multiple objects which can calculate the 6-DOF pose by PPF Voting-Scheme technology. The accuracy of the pose estimation system is obtained from the experiment of pose estimation in which the position error of the linear axis is 1.5mm and the angle error of the rotation axis is within 3 degrees. The third is the intelligent bin-picking system of the robot arm which communicates by the SYNTEC remote API library. We write by SYNTEC PLC and MACRO programs to control the pneumatic gripper I/O and plan the robot arm paths. The system uploads the 6-DOF pose estimation to the robot arm controller which enables the robot arm to grasp a object in the multiple objects and grasp the object to another position. Therefore, the robot arm achieves the function of smart bin-pick. The system enables the robot arm to grasp an individual object in the multiple objects area and grasp this object to the another position. The study achieves the function of smart bin-pick and improves the flexibility of the robot arm on the industrial assembly line to achieve the purpose of unmanned assembly lines.

摘要.....i
Abstract.....ii
誌謝.....iv
目錄.....v
表目錄.....vii
圖目錄.....viii
符號說明.....xi
第 一 章 緒論.....1
1.1 研究背景與動機.....1
1.2 研究目的.....2
1.3 論文架構.....2
1.4 論文貢獻.....2
第 二 章 文獻回顧.....3
2.1 3D視覺應用情境國內外相關研究.....3
2.2 3D視覺輔助機械手臂夾取料件國內外相關研究.....6
第 三 章 研究架構與方法.....12
3.1 研究架構與流程.....12
3.2 硬體及系統架構.....13
3.2.1 機械手臂智慧夾取系統硬體介紹.....13
3.2.2 雷射測距儀硬體設備.....13
3.3 結構光掃描原理與點雲數據處理.....14
3.3.1 投影機虛擬校正板.....14
3.3.2 相機與投影機空間校正原理.....15
3.3.3 結構光編碼與解碼.....19
3.3.4 結構光三角量測原理.....22
3.3.5 統計異常消除濾波器 (Statistical Outlier Removal Filter).....24
3.3.6 網格體素濾波器 ( Voxel Grid Fillter ).....25
3.4 點雲數據特徵點描述演算.....27
3.4.1 表面法向量估計.....27
3.4.2 點對特徵描述(Point Pair Feature, PPF).....28
3.4.3 總體模型描述(Global Model Descriptor).....29
3.4.4 局部座標(Local Coordinates).....30
3.4.5 投票機制演算法(Voting-Scheme Matching).....31
3.4.6 最近點迭代匹配演算法(Iterative Closest Point , ICP).....32
3.5 剛體自由度.....33
3.5.1 剛體自由度.....33
3.6 機械手臂通訊與系統整合建置.....35
3.6.2 機械手臂與相機座標系整合建置.....36
3.6.3 新代控制器通訊系統建置.....36
3.6.4 新代手臂控制器PLC程式.....37
3.6.5 新代MACRO巨集程式控制.....39
第 四 章 實驗結果.....42
4.1 結構光建模與姿態辨識系統建置.....43
4.1.1 結構光建模與姿態辨識平台建置.....43
4.1.2 相機與投影機空間參數校正.....43
4.1.3 資料庫樣本建置流程.....45
4.2 結構光建模與姿態辨識精度實驗.....46
4.2.1 亮度對藍光與白光結構光影響實驗.....46
4.2.2 工件擺放位置對結構光建模精度影響實驗.....48
4.2.3 工件擺放高度對結構光精度影響實驗.....49
4.2.4 估測物件姿態精度實驗.....50
4.3 機械手臂智慧夾取系統建置.....52
4.3.1 機械手臂初始教導與手臂夾取姿態.....52
4.3.2 堆疊物件夾取實驗.....53
4.3.3 機械手臂智慧夾取系統建置.....58
第 五 章 結論與未來展望.....60
參考文獻.....61
Extended Abstract.....63

[1]Rosario Porras-Aguilar, Konstantions Falaggis, Ruben Ramos-Garcia, ”Optimum projection pattern generation for grey-level coded structured light illumination systems”, Volume 91, April 2017, Pages 242-256
[2]Ying He, Bin Liang, Yu Zou, Jin He and Jun Yang, ”Depth Errors Analysis and Correction for Time-of-Flight(TOF) Cameras”, Published: 5 January 2017
[3]Mohammed A. Isa, Danny Sims-Waterhouse, Samanta Piano, Richard Leach, ”Volumetric error modelling of a stereo vision system for error correction in photogrammetric three-dimensional coordinate metrology”, Volume 64, July 2020, Pages 188-199
[4]Stefano Pagano, Riccardo Russo, Sergio Savino, ”A vision guided robotic system for flexible gluing process in the footwear industry”,Volume 65,October 2020.
[5]Eirik B. Njaastad, Olav Egeland, ”Automatic Touch-Up of Welding Paths Using 3D Vision”Volume 49,Issue 31,2016,Pages73-78.
[6]Chengjun Chen, Yong Pan, Dongnian Li, Shilei Zhang, Zhengxu Zhao, Jun Hong,” A virtual-physical collision detection interface for AR-based interactive teaching of robot” Volume 64, August 2020
[7]Lei Wang, Xiaoyun Fan, Jiahao Chen, Jun Cheng, Jun Tan, Xiaoliang Ma, ”3D object detection based on sparse convolution neural network and feature fusion for autonomous driving in smart cities” Volume 54, March 2020, 102002
[8]Yun-Chao Tang, Li-Juan Li, Wen-Xian Feng, Feng Liu, Xiang-Jun Zou, Min-You Chen” Binocular vision measurement and its application in full-field convex deformation of concrete-filled steel tubular columns” Volume 130, December 2018, Pages 372-383
[9]廖寬政，”基於深度學習之輔助取物系統：系統整合及機械手臂取物”，國立台北大學電機工程學系，碩士學位論文，2019。
[10]簡駿宥，”基於物件姿態估測導引機械手臂智慧取放與加工系統”，國立虎尾科技大學，機械與電腦輔助工程系，碩士學位論文，2019。
[11]曾毓書，”任意擺放物件之三維定位與抓取”, 國立雲林科技大學電機工程系，碩士學位論文，2017。
[12]吳宜儒，”基於深度學習與立體影像之機械手臂智慧夾取”，國立清華大學，動力機械工程學系電機控制組，碩士學位論文，2019。
[13]Jay M Wong, Vincent Kee, Tiffany Le, Syler Wagner, Gian-Luca Mariottini, Abraham Schneider, Lei Hamilton, Rahul Chipalkatty, Mitchell Hebert, David Johnson. ”SegICP: Integrated deep semantic segmentation and pose estimation. ”arXiv preprint arXiv:1703.01661, 2017.
[14]Deping Li, Ning Liu, Yulan Guo, Xiaoming Wang, Jin Xu, ”3D object recognition and pose estimation for random bin-picking using Partition Viewpoint Feature Histograms” Volume 128, 1 December 2019, Pages 148-154
[15]Tuan-Tang Le, Chyi-Yeu Lin, ”Bin-Picking for Planar Objects Based on a Deep Learning Network: A Case study of USB Packs” Multidisciplinary Digital Publishing Institute (MDPI) 19 August 2019
[16]Zhengyou Zhang, ”A Flexible New Technique for Camera Calibration”, Technical Report of Microsoft Research, MSR-TR-98-71.
[17]Douglas Lanman, ”Bulid Your Own 3D Scanner: 3D Photography for Begineers”, SIGGRAPH 2009 Course Notes.
[18]Radu Bogdan Rusu,”Semantic 3D Object Maps for Everyday Manipulation in Human Living Environments”, Institut für Informatik , der Technischen Universität München.
[19]Fengjun Hu, Yanwei Zhao,”DISCRETE POINT CLOUD FILTERING AND SEARCHING BASED ON VGSO ALGORITHM”, 27th Conference on Modelling and Sumulation, May 2013.
[20]Bertram Drost, Markus Ulrich, Nassir Navab, Slobodan Ilic,”Model Globally, Match Locally: Efficient and Robust 3D Object Recognition,” in Proc. of IEEE International Conference on Computer Vision and Pattern Recognition(CVPR), Munich, Germany, 2010, pp.998-1005.
[21]Paul J. Besl, Neil D. McKay,” A Method for Registration of 3D Shapes”, IEEE Trans on Pattern Analysis and Machine Intelligence, 1992.
[22]許嘉洧，”基於格雷碼結構光輔助工具機之工件自動定位”，國立虎尾科技大學，機械與電腦輔助工程系，碩士學位論文，2018。
[23]林聖祐，”基於離線編譯之六軸機械手臂銑削路徑規劃”，國立虎尾科技大學，機械與電腦輔助工程系，碩士學位論文，2018。