臺灣博碩士論文加值系統

隨著高人力成本時代來襲，如何減少人力成本變成為現今社會上所需克服議題之一，而本論文所研究題目就是為了減少人力成本損失，並在實驗中規劃出，設定出現今社會中有遇到之問題，並加以解決。以證明本論文之應用性極高。
本論文運用工具機主流加工法『多軸從動』控制邏輯與機器人運動學作為本次自製多軸機械臂主要控制法，透過機器人運動學中的相關知識推導出該手臂的運動方程式，使機械手臂於三維空間座標中自由移動，YOLOv4訓練欲辨識之物件模型，並在測得目標物座標後運用機器人逆向運動學規劃出機械手臂可以移動之範圍，運用Matlab進行運行前移動路徑模擬，用以確認機械臂之規劃路徑是否符合實際規劃路徑，如路徑符合有效範圍內可夾取之路徑，即可將目標物座標轉換成手臂末端坐標至目標物座標間的各軸轉動參數進行手臂移動，運用最短時間路徑規劃出來把所有目標物放置至目標地點。希冀此研究增加工具機及產業之加工效率。

Due to the increment of labor cost, how to reduce labor becomes one of the societal problems that we need to address. The purpose of this paper is to explore and address those societal problems.
This paper uses multi-axis slave and robotic kinematics to control the multi-axis manipulator. We enable the manipulator to move freely in the three-dimensional space. To the end, we derive the equation of motion through robotic kinematics. Moreover, we utilize YOLO v4 to enable this manipulator to recognize the objects so that we can use inverse kinematics to plan the traveling range of the manipulator. To ensure the route is reasonable, we use Matlab to simulate the path the manipulator will move. If the path is reasonble, we can set up the related parameters according to the object’s end coordinates and find the shortest path for the manipulator to move. Hope this research can increase efficiency in the industry.

摘要 I
ABSTRACT II
致謝 III
圖目錄 VI
表索引 VIII
第一章 緒論 1
1.1 前言 1
1.2 研究背景 2
1.3 研究動機與目的 4
1.4 研究論文架構 5
第二章 自製多軸機器臂開發 6
2.1 自製機械臂目標效能 6
2.2 多軸機械手臂機構設計 8
2.2.1 機械臂關節連接設計 9
2.2.2 機械臂負載模擬及分析 10
2.3 機械臂電路控制 12
2.4 機器人運動學 15
2.4.1 機械臂坐標系 16
2.4.2 正向運動學 19
2.4.3 逆向運動學 21
第三章 類神經網路架構 24
3.1 深度類神經網路 24
3.1.1 卷積運算CONVOLUTION 25
3.1.2 激活函數ACTIVATION FUNCTION 25
3.1.3 池化POOLING 26
3.1.4 全連結 FULLY CONNECTED 27
3.2 物件辨識深度神經網路 28
3.3 YOLO V4辨識結果 35
第四章 實驗結果與探討 36
4.1 影像實驗架構 36
4.2 YOLO V4網路訓練架構 37
4.3 MATLAB模擬分析 38
4.4 多軸機械臂實驗結果 40
第五章 結論與未來方向 42
5.1 結論 42
5.2 未來方向 43
參考文獻 44

[1]林沛群, Robotics (1),Coursera ,
Retrieve from: Https://zh-tw.coursera.org/learn/robotics1
[2]蕭俊祥,林顯易, 機器人學,DeltaMOOCx,
Retrieve from:https://univ.deltamoocx.net/
[3]Mark W. Spong, M. Vidyasagar ,Robot Dynamics and Control, 1989.1, Wiley
[4]Document of Robotics Toolbox for Matlab,Peter I. Corke,2001.4,CSIRO
[5]I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning. MIT Press, 2016,
[6]Aston Zhang, Zachary C. Lipton, Mu Li, Alexander J. Smola,Dive into Deep Learning, arXiv:2106.11342
[7]Alexnet: A. Krizhevsky, I. Sutskever, and G. E. Hinton, “Imagenet classification with deep convolutional neural networks,” in Advances in Neural Information Processing Systems, 2012.
[8]Yolov1 J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You Only Look Once: Unified, Real-Time Object Detection,” in 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 779-788.
[9]Yolov2 J. Redmon and A. Farhadi, “YOLO9000: Better, Faster, Stronger,” in 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017, pp. 6517-6525.
[10]Yolov3 J. Redmon and A. Farhadi, “YOLOv3: An Incremental Improvement,” in arXiv e-prints, 2018.
[11]Resnet K. He, X. Zhang, S. Ren, and J. Sun, “Deep Residual Learning for Image Recognition,” in 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 770-778.
[12]YOLOv4 A Bochkovskiy, C Y Wang and H M. Liao, “YOLOv4: Optimal Speed and Accuracy of Object Detection”, 2020.
[13] Kaggle Dataset https://www.kaggle.com/smeschke/four-shapes