臺灣博碩士論文加值系統

在現代的生活中，居住於高樓大廈已是常態，目前在大樓中上下通行的方式，不外乎為樓梯與電梯兩種，而在較多樓層的大廈中，大多數居民的通行方式皆以電梯為主。在這樣的多樓層環境下，傳統的服務型機器人有以下限制，移動範圍僅被限制於單一樓層，無法自主移動到其他樓層，並進行多樓層服務。
以目前市面上的飯店服務型機器人進行分析，有以下的限制，飯店服務型機器人在搭乘電梯進行多樓層移動時，大多是透過WiFi或是藍芽等無線傳輸方式溝通，而為了以無線通訊的方式操作電梯，電梯系統需要重新建置，但重新建置需要考慮到成本以及安全問題。為了解決此問題，我們運用人工智慧影像辨識與機器人作業系統，開發了能自主操作電梯並可全自動於多樓層間移動的服務型機器人，藉以解決此問題。
本文所採用之物體辨識方法為： YOLOv4，而在實驗部分有使用YOLOv3以及YOLOv4進行測試，YOLOv3定位精準、辨識迅速等優勢，而YOLOv4更改了網路架構，使辨識準確度與速度皆有大幅度的提升，而這兩種方法都需事先製作物體標記的訓練樣本，並進行訓練，透過訓練過的模型即可對新影像進行特徵比對與辨識，以達到在未知的環境中尋找物體。
移動機器人的執行平台使用Nvidia Jetson AGX Xavier作為系統核心，並搭配機器人作業系統(ROS)，底盤輪子採用麥克納姆輪使機器人可做全向運動，深度攝影機採用RealSense D435i使機器人擁有深度感知能力，並於移動機器人正前方配置光學雷達YDLIDAR，協助機器人建構地圖以及室內定位。
移動機器人配置以6顆AI馬達所自製之六軸機械手臂，並將夾爪部分設計為多用途以利移動機器人能夠操作電梯介面、按多類型按鈕與取物。當機器人成功接收運送指令後，機器人自主導航與避障前往至指定電梯前，到達目的地後，自主辨識與操作電梯介面，成功操作電梯介面後，移動機器人自主進出電梯。此一研究成果可改善服務型機器人多樓層移動能力，並有潛力可運用於大樓管理、送餐、清潔服務、送貨到府等領域。

The living with high-rising buildings is become normal in modern life. There are usually two ways to move up and down in high-rising buildings, stairs and elevators. It is worth noting that the elevators are often the popular approach for the most of residents. In the multi-storey enviroment, the traditional service robots cannot move to another floor to provide service. However, there are not ready-made approach to provide elevator operating ability for service robot. On the other hand, the elevator could not be modified due to the safe issues.
To solve the issue, a service robot with multi-storey access ability is proposed in this paper. This robot is equipped with a robotic arm that can operate the elevator. Furthermore, the AI object recognition technology is used to detect the buttons of elevators. The detection network use YOLOv4, and the model is trained on a large number of the elevator images, which are collected and labeled by ourselves.
This paper has achieved the goal of service robot with the multi-storey access ability, and can be used to develop robots into food delivery robots, clean service robots.

摘要.....i
Abstract.....ii
誌謝.....iii
目錄.....iv
表目錄.....vii
圖目錄.....viii
第一章 緒論.....1
1.1 研究動機與目的.....1
1.2 問題描述.....1
1.3 研究回顧.....2
1.3.1 即時定位與地圖建構(SLAM).....2
1.3.2 路徑規劃.....3
1.4 電梯面板辨識及操作策略.....3
1.5 論文架構說明.....3
第二章 相關技術分析與文獻探討.....4
2.1 服務型機器人分析與探討.....4
2.2 文獻分析與探討.....4
2.3 影像辨識分析.....6
2.3.1 YOLOv3介紹.....8
2.3.2 YOLOv4介紹.....9
2.4 距離測量分析.....9
2.4.1 超音波測距.....10
2.4.2 紅外線測距.....11
2.4.3 深度測距.....12
2.5 文獻分析.....13
第三章 硬體設計與架構.....14
3.1 硬體設計概述.....14
3.2 機械結構設計.....14
3.3 底盤結構設計.....16
3.4 動力機構設計.....19
3.5 機械手臂設計.....20
3.6 硬體系統架構概述.....23
3.7 機器人系統核心.....24
3.8 影像系統單元.....25
3.8.1 環境偵測系統.....25
3.8.2 影像辨識系統.....27
3.9 底盤驅動單元.....28
3.10 手臂驅動單元.....29
3.11 機器人平行校正與距離計算單元.....30
3.12 電力系統.....31
第四章 軟體系統架構.....34
4.1 深度學習物體辨識.....34
4.1.1 YOLOv3.....34
4.1.2 YOLOv4.....37
4.2 獲取物體距離資訊.....41
4.2.1 影像測距.....41
4.2.2 深度攝影機測距.....42
4.2.3 超音波測距.....43
4.3 SLAM即時定位與地圖構建.....45
4.4 Navigation導航與避障.....46
第五章 實驗方法設計與實現.....48
5.1 跨樓層移動服務型機器人之流程與策略.....48
5.2 AI影像辨識訓練電腦規格與軟體版本.....55
5.2.1 顯示卡資訊.....55
5.2.2 Opencv與CUDA以及CUDNN版本型號.....56
5.2.3 深度學習訓練之電腦規格資訊.....57
5.3 深度學習影像辨識訓練前參數設定.....58
5.3.1 YOLO的訓練流程.....59
5.3.1 準備數據集樣本.....60
5.3.2 標記樣本.....61
5.3.3 將數據進行分割.....62
5.3.4 將XML轉換成VOC格式的txt.....64
5.3.5 生成VOC訓練集、測試集、驗證集的路徑.....65
5.3.6 更改cfg檔案中的參數檔.....66
5.4 YOLOv3與YOLOv4實際訓練過程.....68
5.4.1 第一次訓練.....68
5.4.2 第二次訓練.....70
5.4.3 第三次訓練.....75
5.5 機器人與電梯面板距離偵測.....78
5.5.1 影像測距.....78
5.5.2 深度攝影機測距.....79
5.5.3 超音波測距.....80
5.6 SLAM建構地圖測試.....81
5.7 Navigation導航與避障測試.....83
5.7.2 實驗驗證.....85
第六章 結論與未來展望.....87
6.1 結論.....87
6.2 未來展望.....88
參考文獻.....89

1.Quigley, M., et al. ROS: an open-source Robot Operating System.
2.Redmon, J., et al. You only look once: Unified, real-time object detection. in Proceedings of the IEEE conference on computer vision and pattern recognition. 2016.
3.Grisetti, G., C. Stachniss, and W. Burgard, Improved techniques for grid mapping with rao-blackwellized particle filters. IEEE transactions on Robotics, 2007. 23(1): p. 34.
4.Hart, P.E., N.J. Nilsson, and B. Raphael, A formal basis for the heuristic determination of minimum cost paths. IEEE transactions on Systems Science and Cybernetics, 1968. 4(2): p. 100-107.
5.Koenig, S. and M. Likhachev, Fast replanning for navigation in unknown terrain. IEEE Transactions on Robotics, 2005. 21(3): p. 354-363.
6.Fox, D., W. Burgard, and S. Thrun, The dynamic window approach to collision avoidance. IEEE Robotics & Automation Magazine, 1997. 4(1): p. 23-33.
7.Troniak, D., et al. Charlie Rides the Elevator--Integrating Vision, Navigation and Manipulation towards Multi-floor Robot Locomotion. in 2013 international conference on computer and robot vision. 2013. IEEE.
8.Yin, H., 節能電梯之設計與實現. 2013.
9.Klingbeil, E., et al., Autonomous operation of novel elevators for robot navigation. 2010 IEEE International Conference on Robotics and Automation, 2010: p. 751-758.
10.全球防疫人力吃緊 MSI微星科技AI-UVC主動消毒機器人 Available from:https://www.ankecare.com/2020/18774.
11.佳世達防疫機器人首次亮相 Available from:https://www.ithome.com.tw/news/132596.
12.服務機器人2020年邁入兩位數成長 輔助新冠肺炎防疫風險管理 Available from:https://www2.deloitte.com/tw/tc/pages/technology-media-and-telecommunications/articles/professional-service-robots.html.
13.林晉億, 可自搭電梯之跨樓層自走機器人. 2012.
14.陳漢忠, 智慧型搭乘電梯機器人. 2009.
15.Gatos, B., I. Pratikakis, and S.J. Perantonis, Adaptive degraded document image binarization. Pattern recognition, 2006. 39(3): p. 317-327.
16.An, Q., et al. Elevator button recognition using auto-slant correction and projection histogram. in 2017 10th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI). 2017. IEEE.
17.Zakaria, W.N.F.W., et al., Elevator‘s External Button Recognition and Detection for Vision-based System. Proceeding of the Electrical Engineering Computer Science and Informatics, 2014. 1(1): p. 265-269.
18.Redmon, J. and A. Farhadi, Yolov3: An incremental improvement. arXiv preprint arXiv:1804.02767, 2018.
19.Bochkovskiy, A., C.-Y. Wang, and H.-Y.M. Liao, YOLOv4: Optimal Speed and Accuracy of Object Detection. arXiv preprint arXiv:2004.10934, 2020.
20.Redmon, J. and A. Farhadi. YOLO9000: better, faster, stronger. in Proceedings of the IEEE conference on computer vision and pattern recognition. 2017.
21.He, K., et al. Deep residual learning for image recognition. in Proceedings of the IEEE conference on computer vision and pattern recognition. 2016.
22.Lin, T.-Y., et al. Feature pyramid networks for object detection. in Proceedings of the IEEE conference on computer vision and pattern recognition. 2017.
23.Yun, S., et al. Cutmix: Regularization strategy to train strong classifiers with localizable features. in Proceedings of the IEEE International Conference on Computer Vision. 2019.
24.Ghiasi, G., T.-Y. Lin, and Q.V. Le, Dropblock: A regularization method for convolutional networks. Advances in neural information processing systems, 2018. 31: p. 10727-10737.
25.Wang, C.-Y., et al. CSPNet: A new backbone that can enhance learning capability of cnn. in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. 2020.
26.Benet, G., et al., Using infrared sensors for distance measurement in mobile robots. Robotics and autonomous systems, 2002. 40(4): p. 255-266.
27.Mohammad, T., Using ultrasonic and infrared sensors for distance measurement. World Academy of Science, Engineering and Technology, 2009. 51: p. 293-299.
28.超聲波測距與超聲測距傳感器是這樣玩起來的 Available from:https://kknews.cc/news/6omlx2m.html.
29.兩張圖看懂紅外測距工作原理 Available from:https://kknews.cc/zh-tw/news/a33lvpg.html.
30.3D感測技術發展與應用趨勢 Available from:https://dahetalk.com/2018/03/11/%E3%80%90%E5%9C%96%E8%A7%A3%E3%80%913d%E6%84%9F%E6%B8%AC%E6%8A%80%E8%A1%93%E7%99%BC%E5%B1%95%E8%88%87%E6%87%89%E7%94%A8%E8%B6%A8%E5%8B%A2%EF%BD%9C%E5%A4%A7%E5%92%8C%E6%9C%89%E8%A9%B1%E8%AA%AA/.
31.蘇泓瑞, 具地圖繪製與定位、自主開門之機器人設計與實作. 2018.
32.劉威德, 基於人工智慧與機器人作業系統可影像辨識自主拿取物品之服務型機器人. 2019.
33.YOLOv3 目標檢測演算法詳細總結分析Available from:https://www.itread01.com/content/1543030155.html.
34.He, K., et al., Spatial pyramid pooling in deep convolutional networks for visual recognition. IEEE transactions on pattern analysis and machine intelligence, 2015. 37(9): p. 1904-1916.
35.Liu, S., et al. Path aggregation network for instance segmentation. in Proceedings of the IEEE conference on computer vision and pattern recognition. 2018.
36.Zhang, Z., et al., Bag of freebies for training object detection neural networks. arXiv preprint arXiv:1902.04103, 2019.
37.Müller, R., S. Kornblith, and G.E. Hinton. When does label smoothing help? in Advances in Neural Information Processing Systems. 2019.
38.Lee, H.-C., et al., Probabilistic map merging for multi-robot RBPF-SLAM with unknown initial poses. Robotica, 2012. 30(2): p. 205-220.
39.Navigation Package. Available from: http://wiki.ros.org/navigation.
40.Monte Carlo localization. Available from. Available from: https://en.wikipedia.org/wiki/Monte_Carlo_localization.
41.Thrun, S., W. Burgard, and D. Fox, Probabilistic robotics. 2005: MIT press.
42.Thrun, S., Probabilistic robotics. Communications of the ACM, 2002. 45(3): p. 52-57.