發展應用深度學習之組裝線檢測系統

Development of an Assembly Line Inspection System Using Deep Learning

童晉威(Jin-Wei Tong)

指導教授 ： 陳冠宇

中原大學/工學院/機械工程研究所/碩士(2020年)

https://doi.org/10.6840/cycu202000862

摘要

本文是以樹莓派及個人電腦為基本平台，基於深度學習、機器視覺、圖形使用者介面、機器手臂以及網路傳輸為架構，利用python程式語言進行編譯，發展模擬工廠自動化組裝及檢測系統。實驗前以每秒拍攝一張影像的頻率進行拍攝，藉由影像前處理、影像強化、特徵提取的方式，抓取物件特徵，將處理後的影像作為訓練樣本，搭配設計好的深度學習模型得到實驗參數，經由待測結果物件進行組裝檢測，實驗中使用樹莓派控制機械手臂及旋轉圓盤來進行組裝，由機械手臂夾送待檢測物件至運輸軌道，透過鏡頭進行檢測物件的拍攝，影像讀入事先訓練好的深度學習模型，藉由模型所得到的結果傳輸至樹莓派進行機械手臂或輸送軌道進行動作，並建立一個操作介面可進行遠端即時觀看，本文所發展的組裝線檢測系統可達93%的準確率。

關鍵字

樹莓派 ； 深度學習 ； 卷積神經網路 ； 機器視覺 ； python

並列摘要

This paper uses Raspberry Pi and personal computer as the basic platform, based on deep learning, machine vision, graphical user interface, robot arm and network transmission as the framework, compiles with python programming language, and develops a simulated factory automated assembly and inspection system. Before the experiment, the image was taken at a frequency of one image per second. Through image preprocessing, image enhancement, and feature extraction, the object features were captured, and the processed image was used as a training sample. It was obtained with a designed deep learning model. The experimental parameters are assembled and tested by the object to be tested. In the experiment, the Raspberry Pi is used to control the robotic arm and the rotating disk to assemble. The robotic arm clamps the object to be tested to the transport track. Read the deep learning model trained in advance, and transfer the results obtained by the model to the Raspberry Pi for robotic arm or conveying track for action, and establish an operation interface for remote real-time viewing. The assembly line test developed in this paper The system can reach 93% accuracy.

並列關鍵字

Raspberry Pi ； Deep Learning ； Convolutional Neural Network ； Machine Vision ； Python

參考文獻

[1] D. O. Hebb, The Organization of Behavior. A Neuropsychological Theory. New York.: John Wiley & Sons, Inc, 1949.

[2] F. Rosenblatt,'The Perceptron: A Probabilistic Model for Information Storage and Organization in the Brain,' Cornell Aeronautical Laboratory, Psychological Review, Vol 65, No. 6, pp. 386-408, 1957.

[3] D. Rumelhart, G. Hinton & R. Williams, 'Learning representations by back-propagating errors,' Nature,Vol. 323, pp 533–536, 1986.

[4] G. Hinton, & T . Sejnowski, 'A learning algorithm for boltzmann machines,' Vol. 6088, No. 1, pp 147-169 ,1986.

[5] Y. LeCun, L. Bottou, Y. Bengio & P. Haffner “Gradient-based learning applied to document recognition” Proceedings of the IEEE, Vol. 11, No. 86, pp 2278-2324 1998.