隨者生產模式的轉型，人機協同作業逐漸成為發展的趨勢，然而在這種人機
合作的工作環境之下，人類與機器人之間的距離將大幅度拉近，導致人類與機器
人碰撞的危險性增加，也因此本論文將針對如何促成高效的人機協作環境並同時
兼顧人類與機械手臂的安全進行深入探討。
本研究的目的為即時控制機械手臂以避開工作環境中的人類位置，並在閃避
人類的同時向目標方向移動，藉以促進高效率之人機協作。因此，本研究之主要
工作項目分為三項，首先是對於人機協作環境中人類位置的偵測，而為了近一步
確保工作環境的安全，預測人類短時間內的未來運動軌跡為本研究的第二個重要
項目，最後則為機械手臂之避障路徑規劃，根據環境中之人類位置以及運動軌跡
的預測結果，即時控制機械手臂的運動方向修正，確保人機協作的安全性。而為
了先驗證本研究提出之算法可行性，我們將整體複雜之人機協作環境簡化為二維
平面進行實驗，透過軟體模擬以及實體機械手臂之各項實驗進行驗證。
本研究使用 Kinect 作為環境偵測之感測器，以此感測器獲得工作環境中之人
類位置座標點，並使用樣條回歸模型進行人體運動軌跡的預測，樣條回歸透過其
特有的模型架構，對於人類行為的隨機性與非線性仍能有很好的預測效果，而機
械手臂之避障演算法則採用場效法進行即時避障路徑規劃，且透過本研究提出之
改良方法，成功改善傳統場效法不適用於機械手臂之避障控制的問題，並以人體
運動軌跡預測的結果加入避障控制中，更近一步提高整體安全性。
最後，我們透過 MATLAB 模擬軟體針對提出的各種功能進行模擬驗證，包
括樣條回歸預測結果以及所提出之改良型場效法之可行性，並且將提出之演算法
以實際人機協作環境進行驗證，尋找四位受試者進行實驗，在所有受試者之實驗
數據顯示，本研究所提出之方法在預測精準度與人機協作的安全性皆有很好的表
現，證明本文算法之有效性。

With the modernization of the manufacturing processes, human-machine collaboration has become the trend of future advancement. However, in practical environment of human-machine cooperation, the distance between humans and robots will be shortened, leading to potential danger of humans and robots Therefore, this research explores in depth how to develop an efficient human-robot collaborative environment while taking into account the safety of humans and robotic arms.
The purpose of this research is to control the robotic arm to avoid the human in the working environment and maintain moving towards the target, thereby promoting efficient human-robot collaboration. Therefore, this research divides the main work into three items. The first is the detection of human positions collaborative environment. In order to further ensure the safety, the prediction of human movement trajectories in a short time is the secondpart of this research. The last task is the obstacle avoidance strategy of the robotic arm. According to the predicted results of the human position and prediction, the movement direction of the robotic arm can be adjusted in real time to avoid collision between human and the robot. As a starting plan to verify the feasibility of the proposed ideas, we simplified the human-machine operations to two-dimension plane for experiments, and verified the developed methods through software simulation and various experiments on physical robotic arms.
We use Kinect as the sensor for environmental detection, and use this sensor to obtain the human position, and use the spline regression model to predict the human motion trajectory. The spline regression has a unique model structure that can obtain good predictions even with the randomness and nonlinearity of human behavior. For the obstacle avoidance algorithm of the robotic arm, we use the potential field method for real-time obstacle avoidance planning. The traditional potential field method is not suitable for the obstacle avoidance control of the robot, since the robot is not a moving point. Therefore, the prediction of the human motion trajectory is added to the obstacle avoidance control to further improve the overall safety.
Finally, the MATLAB simulation software is employed to simulate and verify the various proposed algorithms, including the spline regression prediction and the improved potential field method. The overall proposed modifications will be verified in the actual human-machine collaboration environment with physical robot. Four subject operators were recruited to conduct experiments. The experimental data from all the subjects showed that the method proposed in this study has good performance in both prediction accuracy and safety of human-machine collaboration, which support the validity and effectiveness of the algorithms developed in this research.

[1] Hu, N., Wooi-Haw, T., Hau-Lee, T., Junaidi, A. and Ryoichi, K. （2009）. Extraction of human gait features from enhanced human silhouette images. 2009 IEEE International Conference on Signal and Image Processing Applications, pp. 425-430. DOI: 10.1109/ICSIPA.2009.5478691
[2] Du, Y., Fu, Y. and Wang, L. （2016）. Representation Learning of Temporal Dynamics for Skeleton-Based Action Recognition. IEEE Transactions on Image Processing, vol. 25, no. 7, pp. 3010-3022. DOI: 10.1109/TIP.2016.2552404
[3] Luo, R. C., Liao, C. and Kuo, M. （2017）. Non-contact collision avoidance with sensory servo control in real time for industrial automation. 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）, pp. 1-8. DOI: 10.1109/UIC-ATC.2017.8397451
[4] Yu, L., Bai, J. and Ni, C. （2018）. Real-time Perception of Patient Space for Collision Avoidance in Radiation Treatmet. 2018 IEEE-EMBS Conference on Biomedical Engineering and Sciences （IECBES）, pp. 629-632. DOI: 10.1109/IECBES.2018.8626739
[5] Wei, S., Ramakrishna, V., Kanade, T. and Sheikh, Y. （2016）. Convolutional Pose Machines. 2016 IEEE Conference on Computer Vision and Pattern Recognition （CVPR）, pp. 4724-4732. DOI: 10.1109/CVPR.2016.511
[6] Ouyang, W., Chu, X. and Wang, X. （2014）. Multi-source Deep Learning for Human Pose Estimation. 2014 IEEE Conference on Computer Vision and Pattern Recognition, pp. 2337-2344. DOI: 10.1109/CVPR.2014.299
[7] Deng, L. Y., Lim, X. Y. and Ho, S. S. （2021）. Developing A Parser Framework based on OpenPose Skeleton Detection. 2021 IEEE 3rd Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability （ECBIOS）, pp. 165-168. DOI: 10.1109/ECBIOS51820.2021.9510385
[8] Chen, Y., Shen, C., Wei, X., Liu, L. and Yang, J. （2017）. Adversarial PoseNet: A Structure-Aware Convolutional Network for Human Pose Estimation. 2017 IEEE International Conference on Computer Vision （ICCV）, pp. 1221-1230. DOI: 10.1109/ICCV.2017.137
[9] Kohler, M., Using the Kalman filter to track human interactive motion: modelling and initialization of the Kalman filter for translational motion. Citeseer, 1997
[10] Bruce, A. and Gordon, G. （2004）. Better motion prediction for people-tracking. Proc. of the Int. Conf. on Robotics & Automation （ICRA）, Barcelona, Spain
[11] Hasler, N., Stoll, C., Sunkel, M., Rosenhahn, B. and Seidel, H. P. （2009）. A statistical model of human pose and body shape. Computer graphics forum, vol. 28, no. 2, pp. 337-346. Wiley Online Library
[12] Wu, D. and Shao, L. （2014）. Leveraging Hierarchical Parametric Networks for Skeletal Joints Based Action Segmentation and Recognition. 2014 IEEE Conference on Computer Vision and Pattern Recognition, pp. 724-731. DOI: 10.1109/CVPR.2014.98
[13] Mainprice, J., Hayne, R. and Berenson, D. （2015）. Predicting human reaching motion in collaborative tasks using Inverse Optimal Control and iterative re-planning. 2015 IEEE International Conference on Robotics and Automation （ICRA）, pp. 885-892. DOI: 10.1109/ICRA.2015.7139282
[14] Pérez-D'Arpino, C. and Shah, J. A. （2015）. Fast target prediction of human reaching motion for cooperative human-robot manipulation tasks using time series classification. 2015 IEEE International Conference on Robotics and Automation （ICRA）, pp. 6175-6182. DOI: 10.1109/ICRA.2015.7140066
[15] Ghosh, P., Song, J., Aksan, E. and Hilliges, O. （2017）. Learning human motion models for long-term predictions. 2017 International Conference on 3D Vision （3DV）, pp. 458-466. IEEE
[16] Cheng, Y., Zhao, W., Liu, C. and Tomizuka, M. （2019）. Human motion prediction using semi-adaptable neural networks. 2019 American Control Conference （ACC）, pp. 4884-4890. IEEE
[17] Zhao, W., Sun, L., Liu, C. and Tomizuka, M. （2020）. Experimental Evaluation of Human Motion Prediction Toward Safe and Efficient Human Robot Collaboration. 2020 American Control Conference （ACC）, pp. 4349-4354. DOI: 10.23919/ACC45564.2020.9147277
[18] Xu, Z., Liu, X. and Chen, Q. （2019）. Application of Improved Astar Algorithm in Global Path Planning of Unmanned Vehicles. 2019 Chinese Automation Congress （CAC）, pp. 2075-2080. DOI: 10.1109/CAC48633.2019.8996720
[19] Mester, G. （2007）. Obstacle avoidance of mobile robots in unknown environments. 2007 5th International Symposium on Intelligent Systems and Informatics, pp. 123-127. IEEE
[20] Liu, G., Song, C., Zang, X. and Zhao, J. （2018）. Reactive execution of learned tasks with real-time collision avoidance in a dynamic environment. IEEE Access, vol. 6, pp. 57366-57375.
[21] Chen, Z., Ma, L. and Shao, Z. （2019）. Path Planning for Obstacle Avoidance of Manipulators Based on Improved Artificial Potential Field. 2019 Chinese Automation Congress （CAC）, pp. 2991-2996. DOI: 10.1109/CAC48633.2019.8996467
[22] Lu, S. X., Li, E. and Guo, R. （2020）. An Obstacles Avoidance Algorithm Based on Improved Artificial Potential Field. 2020 IEEE International Conference on Mechatronics and Automation （ICMA）, pp. 425-430. DOI: 10.1109/ICMA49215.2020.9233866
[23] Gracia, L., Garelli, F. and Sala, A. （2013）. Reactive Sliding-Mode Algorithm for Collision Avoidance in Robotic Systems. IEEE Transactions on Control Systems Technology, vol. 21, no. 6, pp. 2391-2399. DOI: 10.1109/TCST.2012.2231866
[24] Liu, C. and Tomizuka, M. （2014）. Control in a safe set: Addressing safety in human-robot interactions. Dynamic Systems and Control Conference, vol. 46209, p. V003T42A003. American Society of Mechanical Engineers
[25] Chen, Y., Wang, Y. and Yu, X. （2012）. Obstacle avoidance path planning strategy for robot arm based on fuzzy logic. 2012 12th International Conference on Control Automation Robotics & Vision （ICARCV）, pp. 1648-1653. DOI: 10.1109/ICARCV.2012.6485438
[26] Lin, H., Fan, Y., Tang, T. and Tomizuka, M. （2016）. Human guidance programming on a 6-DoF robot with collision avoidance. 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems （IROS）, pp. 2676-2681. DOI: 10.1109/IROS.2016.7759416
[27] Hamatani, R. and Nakamura, H. （2020）. Collision avoidance control of robot arm considering the shape of the target system. 2020 59th Annual Conference of the Society of Instrument and Control Engineers of Japan （SICE）, pp. 1311-1316. DOI: 10.23919/SICE48898.2020.9240320
[28] Wei, T. and Liu, C. （2019）. Safe Control Algorithms Using Energy Functions: A Uni ed Framework, Benchmark, and New Directions. 2019 IEEE 58th Conference on Decision and Control （CDC）, pp. 238-243. DOI: 10.1109/CDC40024.2019.9029720
[29] Denavit, J. and Hartenberg, R. S. （1955）. A kinematic notation for lower-pair mechanisms based on matrices.
[30] aic1999. （2018）. 机器人理论（3）DH表达法：解析关节轴之间的关系. Available: https://blog.csdn.net/aic1999/article/details/82490615?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522166072955516781432998680%2522%252C%2522scm%2522%253A%252220140713.130102334..%2522%257D&request_id=166072955516781432998680&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2~all~sobaiduend~default-2-82490615-null-null.142^v41^pc_rank_34_1,185^v2^control&utm_term=DH%E8%A1%A8
[31] Zhang, E. （2019）. Kinect 中文教程. Available: https://kinect-tutorials-zh.readthedocs.io/zh_CN/latest/
[32] Funaya, H., Shibata, T., Wada, Y. and Yamanaka, T. （2013）. Accuracy assessment of kinect body tracker in instant posturography for balance disorders. 2013 7th International Symposium on Medical Information and Communication Technology （ISMICT）, pp. 213-217. DOI: 10.1109/ISMICT.2013.6521731
[33] Huber, M., Leeser, M., Sternad, D. and Seitz, A. L. （2015）. Accuracy of kinect for measuring shoulder joint angles in multiple planes of motion. 2015 International Conference on Virtual Rehabilitation （ICVR）, pp. 170-171. DOI: 10.1109/ICVR.2015.7358612
[34] Sabale, A. S. and Vaidya, Y. M. （2016）. Accuracy measurement of depth using Kinect sensor. 2016 Conference on Advances in Signal Processing （CASP）, pp. 155-159. DOI: 10.1109/CASP.2016.7746156
[35] Singh, G. （2018, 20th March）. Introduction to Regression Splines. Available: https://www.analyticsvidhya.com/blog/2018/03/introduction-regression-splines-python-codes/
[36] Khatib, O. （1985）. Real-time obstacle avoidance for manipulators and mobile robots. Proceedings. 1985 IEEE International Conference on Robotics and Automation, vol. 2, pp. 500-505. DOI: 10.1109/ROBOT.1985.1087247