本論文旨在設計一個針對帳單類別信件的自動郵件分揀系統，使用者將多封郵件不分正反面直接放入個別郵件發放裝置，此系統可讓該裝置逐封送出至郵件載台，再透過影像處理定位及辨識該封信件之郵遞區號，隨後使用六自由度機械手臂夾取信件送至對應的郵件分類櫃存放。
本論文之研究項目如下，在硬體部分，透過繪圖軟體及雷射切割機設計並產出零組件而完成以下三點：(1)能逐一發送信封並可暫存多封郵件的個別郵件發放裝置，(2)可供辨識之郵件載台，(3)郵件分類櫃。在郵遞區號辨識演算法的部分，透過郵件載台上下方的兩部網路攝影機完成以下六點：(1)辨識信封正反面，(2)辨識信封傾斜角度並校正，(3)定位及裁切郵遞資訊框，(4)中文字、數字字元分類，(5)利用深度學習網路辨識數字，(6)提取郵遞區號之演算法。另外在機械手臂的運動控制方面，完成以下程序。(1)建置虛擬環境，(2)計算機器手臂運作模型的轉換矩陣，(3)求得從夾取郵件至郵件分類櫃前的各點座標，並以逆運動學控制機械手臂至分類櫃存放信封，(4)最佳化逆運動學移動路徑。綜合上述條件，便可讓系統完成郵遞區號辨識與機械手臂可完成郵件分揀的兩個任務。
本研究在Linux環境下使用機器人作業系統(Robot Operating System, ROS)開發軟體系統，並利用TCP/IP進行資訊傳遞，使多個裝置可進行即時資料傳遞，以實現軟硬體整合協作的設計。本論文在實際實驗中，郵遞區號辨識演算法的實際實驗辨識正確率為95.24%，整體系統的正確分揀率為94.49%，實驗結果顯示本論文確實能成功建立了一套穩定且具有強健性的帳單類別信件自動分揀系統，有助於區域性郵局的日常郵遞業務。

This thesis aims to design an automatic mail sorting system for bill-type letters. This system allows users to place multiple letters into a mail dispensing device, which then sequentially dispatches them to a mail platform. Through image processing, the system recognizes each letter's postal code. It utilizes a 6-DoF robotic arm to transform the letters into corresponding mail-sorting cabinets.
The research objectives of this thesis are as follows: In hardware, designing structure diagrams and manufacture them by a laser cutting machine, the following three issues have been achieved: (1) Mail dispensing device capable of sequentially sending envelope and storing multiple letters, (2) Mail platform for recognition, and (3) Mail sorting cabinet. In the postal code recognition algorithm section, utilizing two web cameras above and below the mail platform, the following six topics were achieved: (1) Identify the front/back side of the bill-type envelope, (2)Recognize and correct the tilt angle of the envelope, (3) Locate and crop the postal information box, (4) Classifying Chinese and numeric characters, (5) Use deep learning network to recognize digits, and (6) Extract the postal code algorithm. Additionally, in the motion control of the robotic arm, the following techniques were completed: (1) Establish a virtual environment, (2) Compute the transformation matrix of the robotic arm operation model, (3) Obtain the coordinates of each point from picking up the mail to storing it in the mail sorting cabinet by inverse kinematic control of the robotic arm, and (4) Optimize the inverse kinematic path-planning. Combining the above conditions allows the system to accomplish postal code recognition and mail sorting by the robotic arm.
This study develops a software system using the Robot Operating System (ROS) in the Linux environment. It utilizes TCP/IP for information transmission, enabling real-time data exchange among multiple devices to achieve the design of software-hardware integration and collaboration. In practical experiments, the postal code recognition algorithm performed a correct recognition rate of 95.24%, and the overall sorting accuracy of the system was 94.49%. The experimental results demonstrate that this thesis has successfully established a stable and robust automatic sorting system for bill-type letters.

[1] 中華郵政股份有限公司：《中華民國111年郵政統計要覽》，中華郵政股份有限公司，2023年6月。
[2] 東芝自動化分揀系統
https://www.businesswire.com/news/home/20180307005451/zh-HK/，2018年3月。
[3] S. Basu, N. Das, R. Sarkar, M. Kundu, M. Nasipuri, and D. K. Basu, "Recognition of Numeric Postal Codes from Multi-script Postal Address Blocks," International Conference on Pattern Recognition and Machine Intelligence, Springer, Berlin, Heidelberg, 2009, pp 381-386.
[4] U. Pal, R. K. Roy, K. Roy, and F. Kimura, "Indian Multi-Script Full Pin-code String Recognition for Postal Automation," 2009 10th International Conference on Document Analysis and Recognition, Barcelona, Spain, 2009, pp. 456-460.
[5] Y. LeCun and C. Cortes, "MNIST handwritten digit database", AT&T Labs, vol. 7, pp. 23, 2010.
[6] W. Jiang, "MNIST-MIX: a multi-language handwritten digit recognition dataset", IOP SciNotes, vol. 1, no. 2, pp. 025002, 2020.
[7] Z. Hongjian, L. Shujing, P. Umapada, and L. Yue, "CNN-Based Hindi Numeral String Recognition for Indian Postal Automation," 2019 International Conference on Document Analysis and Recognition Workshops (ICDARW), Sydney, NSW, Australia, 2019, pp. 77-82.
[8] G. Birjit, P. Sagar, K. Nikhil, D. Shivani, and U. Pooja, "Handwritten Digits Identification Using Mnist Database Via Machine Learning Models," 1st International Conference on Computational Research and Data Analytics (ICCRDA 2020), Rajpura, India, 2020.
[9] E. R. G, S. M, A. R. G, S. D, T. Keerthi, and R. S. R, "MNIST Handwritten Digit Recognition using Machine Learning," 2022 2nd International Conference on Advance Computing and Innovative Technologies in Engineering (ICACITE), Greater Noida, India, 2022, pp. 768-772.
[10] A. K. Agrawal, A. K. Shrivas, and V. k. Awasthi, "A Robust Model for Handwritten Digit Recognition using Machine and Deep Learning Technique," 2021 2nd International Conference for Emerging Technology (INCET), Belagavi, India, 2021, pp. 1-4.
[11] M. Lal Saini, B. Tripathi, and M. S. Mirza, "Evaluating the Performance of Deep Learning Models in Handwritten Digit Recognition," 2023 3rd International Conference on Technological Advancements in Computational Sciences (ICTACS), Tashkent, Uzbekistan, 2023, pp. 116-121.
[12] J. Canny, "A Computational Approach to Edge Detection," in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-8, no. 6, pp. 679-698, Nov. 1986.
[13] Hough, P. V, "Method and means for recognizing complex patterns," U.S. Patent No. 3,069,654, 1962.
[14] C. Zauner, "Implementation and Benchmarking of Perceptual Image Hash Functions, " Hagenberg, Austria, 2010.
[15] D. G. Lowe, "Object recognition from local scale-invariant features," Proceedings of the Seventh IEEE International Conference on Computer Vision, Kerkyra, Greece, 1999, pp. 1150-1157 vol.2.
[16] D. Lowe, "Distinctive Image Features from Scale-Invariant Keypoints," in International Journal of Computer Vision, vol. 60, pp. 91–110, November, 2004.
[17] G. Jocher, A. Chaurasia, and J. Qiu, "Ultralytics YOLO (Version 8.0.0)," https://github.com/ultralytics/ultralytics, 2024.
[18] K. He, X. Zhang, S. Ren, and J. Sun, "Deep residual learning for image recognition", Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 770-778, 2016.
[19] L. Mu, G. Cui, Y. Liu, Y. Cui, L. Fu, and Y. Gejima, "Design and simulation of an integrated end-effector for picking kiwifruit by robot," in Information Processing in Agriculture, vol. 7, pp. 58-71, 2020.
[20] Y. Yu, K. Zhang, H. Liu, L. Yang, and D. Zhang, "Real-Time Visual Localization of the Picking Points for a Ridge-Planting Strawberry Harvesting Robot," in IEEE Access, vol. 8, pp. 116556-116568, 2020.
[21] Y. Xiong, Y. Ge, L. Grimstad, and P. J. From, "An autonomous strawberry‐harvesting robot: Design,development, integration, and field evaluation," in Jounral of Field Robotics, vol. 37, pp. 202-224, 2020.
[22] A. Khan, C. Xiangming, Z. Xingxing, and W. L. Quan, "Closed form inverse kinematics solution for 6-DOF underwater manipulator," in Proc. International Conference on Fluid Power and Mechatronics (FPM), Harbin ,2015, pp. 1171-1176.
[23] J.-J. Kim and J.-J. Lee, "Trajectory optimization with particle swarm optimization for manipulator motion planning," in Proc. IEEE Transactions on Industrial Informatics, vol. 11, pp. 620-631, no, Mar. 2015.
[24] F. Marini and B. Walczak, "Particle swarm optimization (PSO). A tutorial," in Chemometrics and Intelligent Laboratory Systems, vol. 4, part B, pp. 153-165, 2015.
[25] P. Beeson and B. Ames, "TRAC-IK: An open-source library for improved solving of generic inverse kinematics," IEEE-RAS 15th International Conference on Humanoid Robots (Humanoids), Seoul, 2015, pp. 928-935.
[26] S. Kumar, N. Sukavanam, and R. Balasubramanian, "An optimization approach to solve the inverse kinematics of redundant manipulator, " International Journal of Information and System Sciences (Institute for Scientific Computing and Information), vol. 6, no. 4, pp. 414-423, 2010.
[27] J. Vannoy and J. Xiao, "Real-time adaptive motion planning (RAMP) of mobile manipulators in dynamic environments with unforeseen changes," in Proc. IEEE Transactions on Robotics, vol. 24, pp. 1199-1212, Oct. 2008.
[28] J. J. Kuffner Jr and S. M. LaValle, "RRT-Connect: An efficient approach to single-query path planning," in Proc. IEEE International Conference on Robotics and Automation, San Francisco, Aug. 2000, pp. 995-1001.
[29] I. H. Choi and Y. G. Kim, "Head pose and gaze direction tracking for detecting a drowsy driver," International Conference on Big Data and Smart Computing, Bangkok, 2014, pp. 241-244.
[30] J. Denavit and R. S. Hartenberg, "A kinematic notation for lower-pair mechanisms based on matrices," in Journal of Applied Mechanics, vol. 22, pp. 215-221, Jun. 1955.
[31] FLUX Beambox
https://tw.flux3dp.com/beambox/，2024年6月。
[32] 固緯 GPS-3030D 30V 3A 單通道直流電源供應器
https://www.tecpel.com.tw/?model=GPS-3030D#photo0，2024年6月。
[33] Raspberry Pi 4 Model B
https://www.raspberrypi.com/products/raspberry-pi-4-model-b/，2024年6月。
[34] Logitech Webcam C930e
https://www.logitech.com/zh-tw/products/webcams/c930e-business-webcam.960-000976.html，2024年6月。
[35] CAESAR RZ-760 RX 16吋(15.6機身) 144Hz
https://www.cjscope.com.tw/product/detail/117，2024年6月。
[36] 達明機器人
https://www.tm-robot.com/zh-hant/regular-payload/，2024年6月。
[37] CHY2/CHG2 Series
https://www.toyorobot.com/Product/Series/CHG2，2024年6月。
[38] ROS
http://wiki.ros.org/ROS/Tutorials，2024年6月。
[39] MoveIt Concepts
https://moveit.ros.org/documentation/concepts/，2024年6月。
[40] ROS.org rviz
https://wiki.ros.org/rviz，2024年6月。
[41] LabelImg
https://github.com/HumanSignal/labelImg，2022年6月。
[42] 達明機器人
https://assets.omron.eu/downloads/datasheet/en/v3/i837_collaborative_robots_datasheet_en.pdf，2024年6月。
[43] TechmanRobotInc tmr_ros1
https://github.com/TechmanRobotInc/tmr_ros1，2024年1月。