本論文是以樹莓派(Raspberry Pi)做為控制系統，以目前在人工智慧領域較被廣泛使用的Python程式語言編寫控制程式，實現智慧自動化系統。本論文分機械手臂及影像辨識應用、溫度量測、無線網路量測三部份。
機械手臂及影像辨識應用是以鏡頭輸入之影像，經過影像辨識後，讀取目標物位置，再驅動機械手臂前往目標物位置抓取，抓取完成後再依任務需求放置到指定位置。
溫度量測系統之應用是以紅外線溫度感測器量取目標物的溫度，搭配超音波測距感測器量取感測器與目標物的距離，並以一款市面上販售的額溫槍量測目標物的溫度當作正確值來校正實驗值，利用卷積神經網路(CNN)根據輸入的溫度、距離、正確值訓練模型後，成為一套以樹莓派、CNN模型實現的溫度量測校正系統。
無線網路量測系統之應用，使用樹莓派蒐集Wi-Fi強度數據，經過統計處理後，使用循環神經網路Simple RNN模型、LSTM模型預測，針對3個路由器預測訊號的未來變化趨勢。
本論文的量測及處理流程可應用於實現智慧自動化系統。

In third chapter, robotic arm and image recognition, OpenCV is used to complete image processing, HSV color space conversion, and morphological image processing using Gaussian blur, erosion, and dilation to perform a series of input images image processing, we can through the methods mentioned above to eliminate noise of images, and frame each red, green and blue color blocks, and get the center point and angle of the color blocks and mark center into the image.
After obtaining the coordinates of the center point of the color blocks, use the inverse kinematics to calculate the rotation angle of each axis of the six-axis robotic arm, perform the clamping task, and follow the task requirements after clamping the color block place the color block to the designated position, in the study, we can complete the three of tasks which combine with image recognition, color blocks tracking, color block classification and color block stacking.
In fourth chapter, the application of the temperature measurement system, divided into two parts, the fixed distance with temperature change system and the distance change with temperature change system. Both systems use some functions to combine several 4×4 and 6×6 pictures, to put into CNN model.
The fixed distance with temperature change system, we collect 50 data sets, 80% data sets are taken to train CNN model, and 20%taken to predict the accuracy rate, using the relative error method to calculate, take the absolute value and then sum the total to get the average, the average relative error is approximately equal to 1%, the accuracy rate is approximately equal to 99%.
The distance change with temperature change system, we collect 900 data sets, 80% data sets are taken to train CNN model, and 20%taken to predict the accuracy rate, calculate method the same as above, we can get the average relative error is approximately equal to 1%, the accuracy rate is approximately equal to 99%.
In fifth chapter, the application of wireless network measurement system, use Raspberry Pi to measure the Wi-Fi signal strength, use the Python programming language to write the program, make the Raspberry Pi automatically collect data, read the Wi-Fi signal strength every five seconds, for three different routers, read each 201 data at the same time.
According to standard error of the mean (SEM), after taking the average of 7 samples, the standard deviation will be 1/√7 times of the original. Decreasing the sample standard deviation can make the curve smoother. And then, we take seven pieces of data as the input length, as the feature, and the last piece of data as the label. After data processing, we can get 195 data, and use about 80% of data as training data and 20% as test data.
Simple RNN model and LSTM model predict three different routers, the accuracy of the prediction is above 98%, and the predicted trend is very close to the test data.

[1] R. C. Gonzalez and R. E. Woods, Digital Image Processing, Fourth edition, Pearson Education, New York, pp. 635-883,2018.
[2] M. Uzun, L. Ozyilmaz, “Object Motion Detecting Without Color and Object Tracking With Image Processing,” In Journal of Engineering Research and Application ISSN:2248-9622,Vol.9, Issue 3, pp. 17-22, Turkey,2019
[3] Guillermo Guillen, Sensor Projects With Raspberry Pi, Apress, Mexico, pp. 97-139 2019.
[4] B. Jeeva, V. Sanjay, P. Vikram, O. T. Derryl, J. Vinay, “Design and Development of Automated Intelligent Robot Using OpenCV,” In International Conference on Design Innovations for 3Cs Compute Communicate Control (ICDI3C), India, pp.92-96, 2018.
[5] Mark W. Spong, Seth Hutchinson, and M. Vidyasagar, Robot Modeling and Control, First edition, John Wiley & Sons, New York, pp. 65-102, 2007.
[6] Muhammad Aziz Muslim and Seif Nura Urfin, “Design of Geometric-based Inverse Kinematics for a Low Cost Robotic Arm,” In Electrical Power, Electronics, Communications, Controls and Informatics Seminar (EECCIS), Indonesia, pp. 82-86, 2014.
[7] X. Cao, Q. Zhang and Z. Ni, “Design of hybrid visual servo control system of mechanical arm based on image and position,” In IEEE 2nd International Conference on Robotics and Automation Engineering (ICRAE), China, pp. 342-346, 2017.
[8] O. Hock and J. Sedo, “Inverse kinematics using transposition method for robotic arm,” In Proc. ELEKTRO, Slovakia, pp. 1-5, 2018.
[9] Sumardi, L. Febriramadhan and A. Triwiyanto, “Design of color based object sorting through arm manipulator with inverse kinematics method,” In 2016 3rd International Conference on Information Technology, Computer, and Electrical Engineering, Semarang (ICITACEE), Indonesia, pp. 117-122, 2016.
[10] Pu-Sheng Tsai, Ter-Feng Wu, Nien-Tsu Hu and Jen-Yang Chen, “Application of a touch panel for control of a six-axis robotic arm,” In International Conference on Advanced Materials for Science and Engineering (ICAMSE), Taiwan, pp. 640-642, 2016.
[11] R. Muller-Cajar, R. Mukundan, “Triangulation: A new algorithm for Inverse Kinematics,” In Proceedings of Image and Vision Computing New Zealand, New Zealand, pp.181-186, 2007.
[12] D. Kusmenko, K. Schmidt, “Development of an analytical inverse kinematics for a 5 DOF manipulator,” In International Conference on Research and Education in Mechatronics (REM), Germany, 2020.
[13] S. Kucuk, Z. Bingul, “The Inverse Kinematics Solutions of Industrial Robot Manipulators,” In Proceedings of the IEEE International Conference on Mechatronics, Turkey, pp. 274-279, 2004.
[14] M. J. Mnati, R. F. Chisab, A. M. Al-Rawi, A.H. Ali, and A. Van den Bossche, “An open-source non-contact thermometer using low-cost electronic component,” HardwareX, vol. 9, p. e00183, Iraq and Belgium, 2021.
[15] A. Kumar, A. Kumar and M. Kumari, “The Contactless Badge designed IR-Thermometer Evaluting Temperature & Social/Physical Distancing by Fabricating the MLX90614 Thermal Sensor during Covid-19 Global Emergency,” TechRxiv, India, 2021.
[16] I. T. Yuniahastuti, I. Sunaryantiningsih, B. Olanda, “Contactless Thermometer sebagai Upaya Siaga Covid-19 di Universitas PGRI Madiun,” In Jurnal ELECTRA: Electrical Engineering Articles, Vol. 1, Indonesia, pp. 28-34, 2020.
[17] A. Shajkofci, “Correction of human forehead temperature variations measured by non-contact infrared thermometer,” In IEEE Sensors Journal, Switzerland, 2021.
[18] N. Wei-Jie Goh, Jun-Jie Poh, J. Yi Yeo, B. Jun-Jie Aw, Szu Cheng Lai, J. Jian Wei Cheng, C. Yuan Ling Tan, and S. Ken-En Gan, “Design and Development of a Low Cost, Non-Contact Infrared Thermometer with Range Compensation,” Singapore, 2021.
[19] D. Bhardwaj, K. Kataoka and V. Kumar, “Wi-Pi: Distributer Wi-Fi Performance Assessment using Raspberry Pi,” In 2014 on Asian Internet Engineering Conference (AINTEC), India, 2014.
[20] H. Moura, M. Nunes, G. Miranda Jr., D. F. Macedo and L. H. A. Correia, “Estimating Quality of Service on Wi-Fi Stations Using Recurrent Neural Networks” In 2019 IEEE 30th Annual International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC), Brazil, pp. 1-6, 2019.
[21] B. Qolomany, A. Al-Fuqaha, D. Benhaddou and A. Gupta, “Role of Deep LSTM Neural Network and Wi-Fi Networks in Support of Occupancy Prediction in Smart Building,” In Proc. IEEE 19th Int. Conf. High Perform. Comput. Commun, IEEE 15th Int. International Conference on Smart City; IEEE 3rd International Conference on Data Science and Systems (HPCC/SmartCity/DSS), United States of America, pp. 50-57, 2017.
[22] S. Li, Q. Wang, X. Liu and J. Chen, “Low cost LSTM Implementation based on stochastic computing for channel state information prediction,” In IEEE Conferences, IEEE Asia Pacific Conference on Circuits and Systems (APCCAS), China, pp. 231-234, 2017.
[23] X. Yu, L. Xu, L. Ma, Z. Chen and Y. Yan, “Solar radio spectrum classification with LSTM,” In IEEE Conferences, IEEE International Conference on Multimedia & Expo Workshops (ICMEW), Hong Kong, pp. 519-524, 2017.
[24] M.A.A. Haseeb and R. Parasuraman, “Wisture : RNN-based Learning of Wireless Signals for Gesture Recognition in Unmodified Smartphones,” In ArXiv preprint ArXiv:1707.08569. 2017.
[25] P. Zhang, Z. Su and Z. Dong, “Complex Motion Detection Based on Channel State Information and LSTM-RNN,” In Annual Computing and Communication Workshop and Conference (CCWC), United State, 2020.
[26] Matthew A. Carlton and Jay L. Devore, Probability with Applications in Engineering, Science, and Technology, Second edition, Springer Texts in Statistics, United States of America, pp. 88-137, 2017.