在本文中，我們提出了一個仿人類行為的非接觸式實時避碰演算法，可以應用於實驗場域、工廠或是居家服務，提供使用者一個安全且可靠的環境。由於電子計算機有比以往更強大的計算能力，許多以往無法達到實時的演算法，如今都有機會對於動態障礙物進行實時的軌跡規劃，以達到避碰的效果。然而現今的這些演算法，如果套用到我們的類人形機械手臂會顯得相當的奇怪，因為這些演算法所規劃出來的路徑相當的死板，會讓機器人的動作看起來生硬，在一些居家服務的場合，這會使人感到恐懼不安以及不可預測。除此之外，像人類一樣的避碰還有其他好處，例如，人類習慣於尋找最節省能量消耗的動作來移動，模仿人類可以讓我們的系統也節省能量的消耗。因此，我們希望能夠參考人類的動作以及策略來優化現有的演算法，以達到一個類似於人類動作的實時避碰。
首先，我們用深度相機來記錄人類的動作，並將其解耦成對應類人形機械手臂的七個自由度，並訓練成一個預測模型。再來，我們使用了深度感測器來測量環境深度，並且結合人體骨架辨識以及物件辨識來區分機器人看到的東西是障礙物與否，進行障礙物的遮擋判斷。接著，我們結合了零空間控制(Null Space Control) 以及反射向量(Reaction vector)來讓手臂進行即時避碰。最後我們將訓練好的模型，融合目前的實時避碰演算法，並以能量的角度探討表現效能，研究表明人類的資料是的確可以降低機械手臂能量的消耗的。

In this research, we present a human-like non-contact collision avoidance for Humanoid 7-DoF Manipulators. Nowadays because of the better computing abilities of computers more and more algorithms can do the real-time trajectory planning for collision avoidance. However, we are making a home-service dual arms robot, but those existing algorithms won’t avoid obstacles like what human beings do. And there are still some other advantages of acting like a human. For example, humans are used to choosing the energy-saving path. Acting like a human, the system can reduce the consumption of energy. For this reason, we start to program a human-like collision avoidance algorithm. First, we decoupled the 7-DOF of Manipulators into 4-DOF for controlling the position of the tip and 3-DOF for controlling the orientation of the tip. This helped us to map the motion of humans to manipulators. Second, we are looking forward to refer to human motions and combined the Repulsive Reaction Vector Algorithms (RRV) and Multilayer perception Regression (MLP) to design a human-like collision avoidance algorithm. Last, reactive avoidance must be jerky, since the depth sensor is unstable. Therefore, we smooth the obstacle trajectory by using the Savitzky-Golay filter. As a result, we hope that the manipulator can successfully avoid dynamic obstacles in real-time. Experimental results with Dynamixel-arms and Intel Realsense D415 depth sensor.

摘要 I
Abstract II
致謝 III
目錄 IV
圖目錄 VI
表目錄 IX
第一章 緒論 1
1.1 前言 1
1.2 研究動機 1
1.3 文獻回顧 2
1.4 研究目標與方法 5
第二章 類人形七軸機械臂的建模 7
2.1 前言 7
2.2 手臂七自由度的分析與討論 7
2.3 座標系轉換 9
2.3.1 旋轉與轉移矩陣 9
2.3.2 四元素 11
2.4 Denavit-Hartenberg參數 12
2.5 順逆向運動學 14
2.5.1 手臂平面、參考平面以及手臂角度 15
2.5.2 順向運動學 16
2.5.3 逆向運動學 19
2.6 有向包圍盒碰撞偵測方式 21
2.7 本章總結 24
第三章 類人形視覺系統的架設與開法 25
3.1 前言 25
3.2 雙自由度可移動視覺平台 25
3.3 手眼校正 27
3.4 即時障礙物分析 32
3.5 障礙物追蹤、遮擋與交錯 34
3.6 本章總結 37
第四章 人類資料蒐集 38
4.1 前言 38
4.2 人類資料採集裝置 38
4.3 人類資料的解耦 40
4.4 人類資料的預處理 44
4.5 本章總結 45
第五章 類人形即時避免碰撞演算法的開法 46
5.1 前言 46
5.2 類人形七軸手臂軌跡規劃 46
5.3 類人形七軸手臂零空間控制 51
5.4 反射向量應用於即時避碰 54
5.5 仿人避碰模型 57
5.5.1 訓練資料準備 57
5.5.2 多層感知機回歸模型(Multilayer perception, MLP) 58
5.5.3 訓練結果 61
5.6 仿人避碰策略 62
5.7 類人形七軸手臂自碰撞保護 65
5.7.1 角度極限保護 65
5.7.2 身體自碰撞保護 66
5.7.3 腕關節自碰撞保護 68
5.8 仿人避碰實驗 69
5.8.1 實驗架設與評估方式 70
5.8.2 仿人避碰能量耗損實驗 70
5.8.3 腰部彈簧能量耗損實驗 73
5.9 本章總結 74
第六章 結論 75
6.1 總結 75
6.2 本文貢獻 76
6.3 未來展望 77
參考文獻 78

[1] M. Mori, K. F. MacDorman, and N. Kageki, "The Uncanny Valley [From the Field]," IEEE Robotics & Automation Magazine, vol. 19, no. 2, pp. 98-100, 2012, doi: 10.1109/MRA.2012.2192811.
[2] A. Bicchi and G. Tonietti, "Fast and "soft-arm" tactics [robot arm design]," IEEE Robotics & Automation Magazine, vol. 11, no. 2, pp. 22-33, 2004, doi: 10.1109/MRA.2004.1310939.
[3] S. Khan, G. Herrmann, T. Pipe, C. Melhuish, and A. Spiers, "Safe Adaptive Compliance Control of a Humanoid Robotic Arm with Anti-Windup Compensation and Posture Control," International Journal of Social Robotics, vol. 2, pp. 305-319, 09/01 2010, doi: 10.1007/s12369-010-0058-7.
[4] M. W. Strohmayr, H. Wörn, and G. Hirzinger, "The DLR artificial skin step I: Uniting sensitivity and collision tolerance," in 2013 IEEE International Conference on Robotics and Automation, 6-10 May 2013 2013, pp. 1012-1018, doi: 10.1109/ICRA.2013.6630697.
[5] R. Dahiya, "E-Skin: From Humanoids to Humans [Point of View]," Proceedings of the IEEE, vol. 107, no. 2, pp. 247-252, 2019, doi: 10.1109/JPROC.2018.2890729.
[6] B. R. Donald, "A search algorithm for motion planning with six degrees of freedom," Artificial Intelligence, vol. 31, no. 3, pp. 295-353, 1987/03/01/ 1987, doi: https://doi.org/10.1016/0004-3702(87)90069-5.
[7] R. Bohlin and L. E. Kavraki, "Path planning using lazy PRM," in Proceedings 2000 ICRA. Millennium Conference. IEEE International Conference on Robotics and Automation. Symposia Proceedings (Cat. No.00CH37065), 24-28 April 2000 2000, vol. 1, pp. 521-528 vol.1, doi: 10.1109/ROBOT.2000.844107.
[8] T. Tao, X. Zheng, H. He, J. Xu, and B. He, "An Improved RRT Algorithm for The Motion Planning of Robot Manipulator Picking up Scattered Piston," in 2018 IEEE 4th Information Technology and Mechatronics Engineering Conference (ITOEC), 14-16 Dec. 2018 2018, pp. 234-239, doi: 10.1109/ITOEC.2018.8740527.
[9] M. Shimizu, H. Kakuya, W. Yoon, K. Kitagaki, and K. Kosuge, "Analytical Inverse Kinematic Computation for 7-DOF Redundant Manipulators With Joint Limits and Its Application to Redundancy Resolution," IEEE Transactions on Robotics, vol. 24, no. 5, pp. 1131-1142, 2008, doi: 10.1109/TRO.2008.2003266.
[10] Y. Qian and A. Rahmani, "Path Planning Approach for Redundant Manipulator Based on Jacobian Pseudoinverse-RRT Algorithm," in Intelligent Robotics and Applications, Berlin, Heidelberg, J. Lee, M. C. Lee, H. Liu, and J.-H. Ryu, Eds., 2013// 2013: Springer Berlin Heidelberg, pp. 706-717.
[11] M. Wang, J. Luo, and U. Walter, "A non-linear model predictive controller with obstacle avoidance for a space robot," Advances in Space Research, vol. 57, no. 8, pp. 1737-1746, 2016/04/15/ 2016, doi: https://doi.org/10.1016/j.asr.2015.06.012.
[12] J. van den Berg, D. Ferguson, and J. Kuffner, Anytime Path Planning and Replanning in Dynamic Environments. 2006, pp. 2366-2371.
[13] Web of Science, "collision avoidance real time." [Online]. Available: http://apps.webofknowledge.com/CitationReport.do?product=WOS&search_mode=CitationReport&SID=E6uMQQV55scruCfaZMw&page=1&cr_pqid=3&viewType=summary&colName=WOS.
[14] S. Haddadin et al., "Real-time reactive motion generation based on variable attractor dynamics and shaped velocities," in 2010 IEEE/RSJ International Conference on Intelligent Robots and Systems, 18-22 Oct. 2010 2010, pp. 3109-3116, doi: 10.1109/IROS.2010.5650246.
[15] D. Han, H. Nie, J. Chen, and M. Chen, "Dynamic obstacle avoidance for manipulators using distance calculation and discrete detection," Robotics and Computer-Integrated Manufacturing, vol. 49, pp. 98-104, 2018/02/01/ 2018, doi: https://doi.org/10.1016/j.rcim.2017.05.013.
[16] R. C. Luo, M. Ko, Y. Chung, and R. Chatila, "Repulsive reaction vector generator for whole-arm collision avoidance of 7-DoF redundant robot manipulator," in 2014 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, 8-11 July 2014 2014, pp. 1036-1041, doi: 10.1109/AIM.2014.6878217.
[17] R. C. Luo, C. Liao, and M. Kuo, "Non-contact collision avoidance with sensory servo control in real time for industrial automation," in 2017 IEEE SmartWorld, Ubiquitous Intelligence & Computing, Advanced & Trusted Computed, Scalable Computing & Communications, Cloud & Big Data Computing, Internet of People and Smart City Innovation (SmartWorld/SCALCOM/UIC/ATC/CBDCom/IOP/SCI), 4-8 Aug. 2017 2017, pp. 1-8, doi: 10.1109/UIC-ATC.2017.8397451.
[18] F. Flacco, T. Kroger, A. Luca, and O. Khatib, "Depth space approach to human-robot collision avoidance," Proceedings - IEEE International Conference on Robotics and Automation, pp. 338-345, 05/01 2012, doi: 10.1109/ICRA.2012.6225245.
[19] S. Kim, E. Gribovskaya, and A. Billard, "Learning Motion Dynamics to Catch a Moving Object," 2010 10th IEEE-RAS International Conference on Humanoid Robots, Humanoids 2010, 12/01 2010, doi: 10.1109/ICHR.2010.5686332.
[20] A. Bochkovskiy, C.-Y. Wang, and H.-y. Liao, YOLOv4: Optimal Speed and Accuracy of Object Detection. 2020.
[21] Z. Cao, G. Hidalgo, T. Simon, S.-E. Wei, and Y. Sheikh, OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields. 2018.
[22] S. Gottschalk, M. Lin, and D. Manocha, "OBBTree: A Hierarchical Structure for Rapid Interference Detection," Computer Graphics, vol. 30, 10/04 1997, doi: 10.1145/237170.237244.
[23] R. S. Hartenberg and J. Denavit, "A kinematic notation for lower pair mechanisms based on matrices," Journal of applied mechanics, vol. 77, no. 2, pp. 215-221, 1955.
[24] Pushpendra050, "Denavit–Hartenberg parameters." [Online]. Available: https://commons.wikimedia.org/w/index.php?curid=87693685.
[25] Johnny Huynh, "Separating Axis Theorem for Oriented Bounding Boxes." [Online]. Available: https://www.jkh.me/files/tutorials/Separating%20Axis%20Theorem%20for%20Oriented%20Bounding%20Boxes.pdf.
[26] C. Ericson, Real-Time Collision Detection. CRC Press, Inc., 2004, p. 632.
[27] TECHNOCHIC, "MOVEMENT & PLANES OF MOTION." [Online]. Available: https://technochicblog.wordpress.com/2013/02/12/movement-planes-of-motion/.
[28] 採智科技股份有限公司, "PRODUCT DESCRIPTION." [Online]. Available: https://idminer.com.tw/product/dynamixel-mx-%E7%B3%BB%E5%88%97%E5%85%A8%E5%90%91%E6%99%BA%E8%83%BD%E9%A6%AC%E9%81%94%E7%89%B9%E8%89%B2%E8%AA%AA%E6%98%8E/.
[29] I. Realsense, "DEPTH CAMERA D435." [Online]. Available: https://www.intelrealsense.com/depth-camera-d435/.
[30] tony2278, "Hand-Eye-Calibration-Matlab." [Online]. Available: https://www.twblogs.net/a/5c2b26e4bd9eee01606d6073.
[31] E. Olson, AprilTag: A robust and flexible visual fiducial system. 2011, pp. 3400-3407.
[32] K. S. Arun, T. S. Huang, and S. D. Blostein, "Least-Squares Fitting of Two 3-D Point Sets," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-9, no. 5, pp. 698-700, 1987, doi: 10.1109/TPAMI.1987.4767965.
[33] V. Klema and A. Laub, "The singular value decomposition: Its computation and some applications," IEEE Transactions on Automatic Control, vol. 25, no. 2, pp. 164-176, 1980, doi: 10.1109/TAC.1980.1102314.
[34] A. Bewley, Z. Ge, L. Ott, F. Ramos, and B. Upcroft, "Simple Online and Realtime Tracking," 02/01 2016.
[35] F. Chun-Hao and C. Jen-Yuan, "A 3D Multi-Object Tracking Based on Bounding Box and Depth," in Information Storage and Processing Systems, 2021: American Society of Mechanical Engineers.
[36] N. Gallagher, Savitzky-Golay Smoothing and Differentiation Filter. 2020.
[37] 鄭期元 and 胡竹生, "Motion Trajectory Estimation on Earth Surface by
Integrating GNSS and INS Measurements," Master dissertation,. p. 31, 2013.
[38] xiaozhuchacha, "Kinect2Toolbox," 2017. [Online]. Available: https://github.com/xiaozhuchacha/Kinect2Toolbox.
[39] J. J. Kuffner and S. M. LaValle, "RRT-connect: An efficient approach to single-query path planning," in Proceedings 2000 ICRA. Millennium Conference. IEEE International Conference on Robotics and Automation. Symposia Proceedings (Cat. No.00CH37065), 24-28 April 2000 2000, vol. 2, pp. 995-1001 vol.2, doi: 10.1109/ROBOT.2000.844730.