近十年來，移動式機械臂的發展已越來越多樣，不只設計來執行抓取任務，還可以用來執行組裝任務，同時對於能夠執行插梢裝配之機械手臂的要求也隨之提升。本研究採用實驗室自行研發之眼在手移動式機械手臂做為研發平台，其由七軸六自由度之機械手臂及全方位移動平台組成。在執行插梢裝配任務時，首先利用視覺伺服算法控制手臂末端之插梢棒接近目標孔，然後利用模糊滑動阻抗控制繼續將插梢棒插入孔中。本論文之重點在於應用深度卷積神經網路在插梢的第一階段尋找插梢孔之影像特徵。為了提高對於光反射及背景雜物的強健性，不僅採用了 Unet模型，還提出了一種基於連通成分分析的影像前處理方法。最後進行了一組實驗，以驗證移動式機械手臂在公差為 0.79mm 的情況下進行插梢裝配任務之效果。

In the recent decade, mobile manipulators have been designed not only to execute pick-andplace tasks, but also to perform assembly tasks. There is increasing demand for mobile manipulators capable of performing peg-in-hole insertion tasks. An eye-in-hand mobile manipulator that was constructed in our laboratory is adopted as a research platform. When performing the assembly task, the mobile manipulator holding a peg is first controlled to approach a target hole by a visual servoing algorithm. Then, the manipulator continues to insert the peg into the hole by a fuzzy sliding impedance control algorithm. The focus of this thesis is to apply a deep convolutional neural network to find the image features of the hole in the first stage of motion. To improve the robustness to specular reflections and cluttered scenes, not only is the model of Unet employed, but also an image preprocessing approach, based on connected component analysis, is proposed. Finally, a set of experiments are conducted to verify the effectiveness of peg-in-hole insertion task with a tolerance of 0.79 mm by a mobile manipulator.

[1] S. Bogh, M. Hvilshoj, M. Kristiansen , and O. Madsen, “Autonomous industrial mobile manipulation (AIMM): from research to industry,” Industrial Robot: the international journal of robotics research and application, vol. 12, no. 2,pp. 120-135, 2012.
[2] T. Kalsoom, N. Ramzan, S. Ahmed, and M. Ur-Rehman, “Advances in sensor technologies in the era of smart factory and industry 4.0,” Sensors, vol. 20, Issue. 23, pp. 6783, 2020.
[3] C. C. Lin, and M. Y. Chen, "Intelligent mobile mechanical arm design.” 2010 International Conference on System Science and Engineering. IEEE, pp. 136-141, 2010.
[4] A. Dunkels, "Design and Implementation of the lwIP TCP/IP Stack." Swedish Institute of Computer Science, vol. 2, No. 77, 2001.
[5] J.F. Broenink, and M. L.J. Tiernego, “Peg-in-Hole assembly using Impedance Control with a 6 DOF Robot,“ Proceedings of the 8th European Simulation Symposium, pp. 504-508, 1996.
[6] K. Fukushima, S. Miyake, “ Neocognitron: A self-organizing neural network model for a mechanism of visual pattern recognition,” Competition and cooperation in neural nets, pp. 267-285, 1982.
[7] Y. LeCun, Y. Bengio, “Convolutional networks for images, speech, and time series,” The handbook of brain theory and neural networks 3361.10, 1995.
[8] F. Yu, V. Koltun, “ Multi-scale context aggregation by dilated convolutions, “ arXiv preprint arXiv:1511.07122 , 2015.
[9] V. Badrinarayanan, A. Kendall, and R. Cipolla, “Segnet: A deep convolutional encoder-decoder architecture for image segmentation,” IEEE transactions on pattern analysis and machine intelligence, vol. 39, Issue. 12, pp. 2481-2495, 2017.
[10] Y. Wei, H. Xiao, H. Shi, Z. Jie, J. Feng, and T. S. Huang, “Revisiting dilated convolution: A simple approach for weakly-and semi-supervised semantic segmentation,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 7268-7277, 2018.
[11] W. Kim, and J. Seok, “Indoor semantic segmentation for robot navigating on mobile,” 2018 Tenth International Conference on Ubiquitous and Future Networks (ICUFN), pp. 22-25, 2018.
[12] J. Long, E. Shelhamer, and T. Darrel, “ Fully convolutional networks for semantic segmentation,” Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 3431–3440, 2015.
[13] J. Dai, Y. Li, K. He, & J. Sun, “R-fcn: Object detection via region-based fully convolutional networks, “ In Advances in neural information processing systems, pp. 379-387.
[14] O. Ronneberger, P. Fischer, and T. Brox, “ U-Net: convolutional networks for biomedical image segmentation,” Medical Image Computing and Computer-Assisted Intervention, pp. 234–241, 2015.
[15] X. Li, H. Chen, X. Qi, Q. Dou, C. W. Fu, and P. A. Heng, “H-DenseUNet: hybrid densely connected UNet for liver and tumor segmentation from CT volumes,” IEEE transactions on medical imaging, vol. 37, Issue. 12, pp. 2663-2674, 2018.
[16] Y. Weng, T. Zhou, Y. Li, & X. Qiu, “Nas-unet: Neural architecture search for medical image segmentation, ” IEEE Access, vol. 7, pp. 44247-44257, 2019.
[17] S. Guan, A. A. Khan, S. Sikdar, & P. V. Chitnis, “Fully dense UNet for 2-D sparse photoacoustic tomography artifact removal, ” IEEE journal of biomedical and health informatics, vol. 24, Issue. 2, pp. 568-576, 2019.
[18] N. Enshaei, S. Ahmad, and F. Naderkhani, “Automated detection of textured-surface defects using UNet-based semantic segmentation network,” 2020 IEEE International Conference on Prognostics and Health Management (ICPHM), pp. 1-5, 2020.
[19] P. C. Yang, K. Sasaki, K. Suzuki, K. Kase, S. Sugano, & T. Ogata, “Repeatable folding task by humanoid robot worker using deep learning, ” IEEE Robotics and Automation Letters, vol. 2, Issue. 2, pp. 397-403, 2016.
[20] A. Anglani, F. Taurisano, R. De Giuseppe, and C. Distante, “Learning to Grasp by using Visual Information,” Proceedings of the IEEE International Symposium on Computational Intelligence in Robotics and Automation, pp. 7-14, Nov. 1999.
[21] R. D. Giuseppe, F. Taursano, C. Distante, and A. Anglani, “Visual servoing of a robotic manipulator based on fuzzy logic control,” Proceedings of the IEEE International Conference on Robotics and Automation, Vol. 2, pp. 1487-1494, May. 1999.
[22] Z. Wasik and A. Saffiotti, “A hierarchical behavior-based approach to manipulation tasks,” Proceedings of the IEEE International Conference on Robotics and Automation, pp. 2780-2785, Set. 2003.
[23] Z. Wasik and A. Saffiotti, “A fuzzy behavior-based control system for manipulation,” Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 1596-1601, Oct. 2002.
[24] H. S. Kim and J. B. Song, “Multi-DOF counterbalance mechanism for a service robot arm,” IEEE/ASME Transactions on Mechatronics, Vol. 19, No. 6, pp. 1756-1763, Dec. 2014.
[25] H. Wang et al. “Research on the Relationship between Classic Denavit-Hartenberg and Modified Denavit-Hartenberg,” International Symposium on Computational Intelligence and Design, Dec. 2014
[26] W. Khalil and J. F. Kleinfinger, “A new geometric notation for open and closed loop robots,” Proceedings of the IEEE International Conference on Robotics and Automation, pp. 1174-1179, 1986.
[27] R. P. Paul, “Robot manipulators: mathematics, programming, and control: the computer control of robot manipulators. ” Richard Paul, 1981.
[28] J. W. Lin, Intelligent Control of Eye-in-hand Mobile Manipulators for Peg-in-hold Assembly, M. S. Thesis, Department of Mechanical Engineering, National Cheng Kung, R.O.C., 2020.
[29] H. mai, M. Iri, Computational-geometric Methods for Polygonal Approximations of a Curve Comput. Vision Graph. Image Process , pp. 31-41, 1986
[30] P. J. Hung, Contruction of an Omni-directional Mobile Manipulator abd Its Behavioristic Image-based Pose Control Using Eye-in-hand Vision System. M. S. Thesis, Department of Mechanical Engineering, National Cheng Kung, R.O.C., 2014.
[31] Y. I. Liu, Construction and Intelligent Control of a Feeding Humanoid Robot and Its Application, M. S. Thesis, Department of Mechanical Engineering, National Cheng Kung, R.O.C., 2010.
[32] W. J. Liao, Intelligent Impedance Control of Robot Manipulators for Peg-in-hold Assembly, M. S. Thesis, Department of Mechanical Engineering, National Cheng Kung, R.O.C., 2019.