在木造建築和室內裝潢行業中，大量地使用了氣動與電動釘槍 (Nailing Gun) 與各種不同種類的緊固件(Fastener)，或稱釘子作為結合建材的工具。而釘子擊發後的異常狀態，包括浮釘、滑釘或者誤用釘型，都有機會對人體造成嚴重傷害，更甚者危急居住安全。不熟悉工具、釘子本身品質較差或操作人員粗心都可能導致異常狀態。然而，傳統的釘子異常檢測方法依賴於用戶的經驗，進而產生誤判或者忽略異常的可能，原因通常是由環境亮度、工作效率要求或注意力不集中等多種原因造成的。本研究提出了一種從工具外部、獨立於使用者經驗的系統來解決上述問題，主要利用人工智慧和自主移動機器人，在悲劇發生之前，儘早發現風險並警告操作者。本研究使用空中機器人檢測屋頂和牆壁較高位置的釘子狀態，而地面移動機器人則負責識別室內裝潢和工作空間較低的位置。這些由機器人收集到的圖像被傳輸到具有預先訓練好的深度學習模型的系統，以識別釘子類別和狀態，確認是否發生異常。在本篇論文中比較了不同的深度學習算法，包括 Faster R-CNN、 Single Shot multi-box Detection 和 You Only Look Once，然後為系統選擇最合適的模型。

It is known that in construction and decorating industry, an abnormal fastening status, including proud nail, mis-fired or mis-used fastener will lead serious injury to human body and jeopardize the resident safety. An abnormal status can be caused by unfamiliarity with the tool, poor quality of the fastener or operator careless. However, the conventional detecting method to fastening failure relies on users’ experience, which ignorance usually occurs by multiple reasons such as environment brightness, working efficient requirement or decreasing of concentration. This research proposed a system from outside of the fastener driving equipment and independent to user experience to solve the problem mentioned above. The aim of this study is to use artificial intelligence and mobile robot to detect the risk and
alarm the operator in early stage before the tragedy happens. The ariel mobile robot collects the data and detects the fastener status on the roof and higher position of the wall while the ground mobile robot identifies the issues for interior decoration and lower position of the working space. The image is transferred to a system with pre-trained deep learning model to identify the fastener category and status. In this research it compares different deep learning algorithm including Convolutional Neural Network, Faster Regional Convolutional Neural Network, Single shot multi-box detection and You Only Look Once then choose the most suitable model for the system. The research concluded the Single shot multi-box detection performed best with highest mean average precision and lower log-average miss rate at fastener detection.

[1] C. A. Holcroft, and L. Punnet, “Work environment risk factors for injuries in wood processing,” Journal of Safety Research, vol.40, no. 4, pp. 247-255, Aug. 2009.
[2] H. J. Lipscomb, A. L. Schoenfisch, and K. S. Shishlov, “Non-fatal contact injuries among workers in the construction industry treated in US emergency departments. 1998–2005,” Journal of Safety Research, vol.41, no. 3, pp. 191-195, Jun. 2010.
[3] W. F. Wen, “Pneumatic nail gun,” U.S. Patent 7 290 691, Oct. 19, 2006.
[4] I. T. Wu, H. L. Ma, and Z. L. Liao, “Pneumatic nail gun capable of striking nails in automatic mode,” U.S. Patent 10 016 884, Jul. 10, 2018.
[5] C. S. Liang and C. H. Tseng, “Actuator for electrical nail gun,” U.S. Patent 7 575 141, Feb. 4, 2008.
[6] R. L. Leimbach, T. A. McCardle, D. L. Bolender, S. Dickinson, J. R. Knueven, R. L. Lance, D. Stoltz, and M. V. Petrocelli, “Fastener driving tool using a gas spring,” U.S. Patent 11 034 007, Jun. 15, 2021.
[7] N. Dalal and B. Triggs, “Histograms of Oriented Gradients for Human Detection,” in Proc. IEEE Conference on Computer Vision and Pattern Recognition, San Diego, CA, USA, 2005, pp.886–893.
[8] J. Uijlings, K. Sande, T. Gevers, and A.W. M. Smeulders, “Selective Search for Object Recognition,” International Journal of Computer Vision, vol.104, pp. 154-171, Mar. 11, 2013.
[9] C. L. Zitnick and P. Doll´ar, “Edge boxes: Locating object proposals from edges,” in Proc. European Conference on Computer Vision, Zurich, Switzerland, 2014, pp. 391–405.
[10] D. G. Lowe, “Distinctive Image Features from Scale-Invariant Keypoints,” International Journal of Computer Vision, vol.60, no.2, pp. 91–110, Jan. 5, 2004.
[11] O. Deniz, G. Bueno, J. Salido, and F. De la Torre, “Face recognition using histograms of oriented gradients,” PR Letters, vol. 32, no. 12, pp. 1598–1603, Sep. 1, 2011.
[12] R. Girshick, J. Donahue, T. Darrell, and J. Malik, “Rich feature hierarchies for accurate object detection and semantic segmentation,” in Proc. IEEE Conference on Computer Vision and Pattern Recognition, Columbus, OH, USA, 2014, pp. 580-587.
[13] R. Girshick, “Fast R-CNN,” in Proc. IEEE International Conference on Computer Vision, Santiago, Chile, 2015, pp.1440-1448.
[14] S. Ren, K. He, R. Girshick and J. Sun, "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 39, no. 6, pp. 1137-1149, Jun. 1, 2017.
[15] J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You Only Look Once: Unified, Real-Time Object Detection,” in Proc. IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, Nevada, USA, 2016, pp. 779-788.
[16] W. Liu et al., “SSD: Single Shot MultiBox Detector,” in Proc. European Conference on Computer Vision, Amsterdam, Netherlands, 2016, pp.21-37.
[17] J. Redmon and A. Farhadi, "YOLO9000: Better, Faster, Stronger," in Proc. IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, USA, 2017, pp. 6517-6525.
[18] J. Redmon and A. Farhadi, “Yolov3: An incremental improvement,” arXiv preprint arXiv:1804.02767, 2018.
[19] I. Goodfellow, Y. Bengio, and A. Courville, “Convolutional Networks” in Deep Learning, Boston, USA: MIT Press, 2016, pp. 326-366.
[20] L. Deng and D. Yu, "Deep Learning: Methods and Applications," Foundations and Trends® in Signal Processing, vol. 7, no. 3–4, pp. 197-387, Jun. 30, 2014.
[21] H. Lee, R. Grosse, R. Ranganath, and A.Y. Ng, “Convolutional deep belief networks for scalable unsupervised learning of hierarchical representations,” in Proc. Annual International Conference on Machine Learning, Montreal, Quebec, Canada, 2009, pp. 609–616.
[22] H. Voos, “Nonlinear state-dependent Riccati equation control of aquadrotor UAV,” in Proc. IEEE International Conference on Control Applications, Munich, Germany, 2006, pp 2547-2552.
[23] A. Cabarbaye, R. L. Leal, P. Fabiani and M. B. Estrada, "VTOL aircraft concept, suitable for unmanned applications, with equivalent performance compared to conventional aeroplane," in Proc. International Conference on Unmanned Aircraft Systems, Arlington, VA, USA, 2016, pp. 219-226.
[24] A. Tayebi and S. McGilvray, “Attitude stabilization of a VTOL quadrotor aircraft,” IEEE Transaction on Control System Technology, vol.14, no.3, pp. 562 – 571, May 2006.
[25] T. Bresciani, “Modelling, Identification and Control of a Quadrotor Helicopter,” Ph.D. dissertation, Lund University, Lund, ‎Scania‎, Sweden‎, 2008.
[26] D. Lee, H. J. Kim, and S. Sastry, “Feedback linearization vs. adaptive sliding mode control for a quadrotor helicopter,” International Journal of Control, Automation and Systems, vol. 7, no. 3, pp. 419–428, May. 30, 2009.
[27] F. Sharifi, M. Mirzaei, B. Gordon, and Y. Zhang, “Fault tolerant control of a quadrotor uav using sliding mode control,” in Proc. Conference on Control and Fault-Tolerant Systems, Nice, France, 2010, pp. 239 –244.
[28] P. Doll´ar, C. Wojek, B. Schiele, and P. Perona, “Pedestrian detection: An evaluation of the state of the art,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol.34, no.4, pp.743-761, April 2012.
[29] G. Salton and M. J. McGill, ”Retrieval Evaluation” in Introduction to Modern Information Retrieval, New York, USA: McGraw-Hill, Inc., 1986, pp. 157-191.