數位影像相關法於精密量測與人機共工系統的整合應用

Integration of Digital Image Correlation Method for Precise Measurement and Human-Robot Collaboration System

謝佳軒(Chia-Hsuan Hsieh)

指導教授 ： 馬劍清

國立臺灣大學/工學院/機械工程學研究所/碩士(2022年)

https://doi.org/10.6342/NTU202203028

摘要

數位影像相關法(Digital image correlation, DIC)是一種非接觸且全域性的光學量測技術，可應用於跨尺度與跨領域的工程問題上，透過影像追蹤待測物表面特徵，可量測具有次像素的位移，資料再經數值微分可獲得速度、加速度及應變等多種物理量。本文首先將數位影像相關法應用在懸臂板受力及受熱變形的精密全場位移量測，並以雷射位移計與有限元素法驗證受力變形的量測結果，熱變形的量測結果則可以反算材料的熱膨脹係數，與理論值互相驗證。
數位影像相關法亦可應用於智慧機械的量測，以開發中的機械手臂為測試對象，結合高速攝影機分析機械手臂於高速運動的位移、振動、速度及加速度，評估機械手臂在開發階段可能存在的系統問題。由機械手臂關節上貼附的特徵點，可監控機械手臂的運動軌跡及各軸關節角度，並還原機械手臂姿態達到虛實整合的效果。對於機械手臂周圍動件的監測，系統整合OpenPose、影像矩不變量和樣板比對辨識，並和其他實驗室開發之即時光學投影系統、人體動作辨識系統、手臂避障路徑規劃系統完成整合，模擬實際工業產線的應用情境。整合的系統以全影像的方式進行量測，監測機械手臂與周圍障礙物以及工作人員之間的最短距離，當距離小於使用者設定安全值時，系統會發出警示提醒工作人員，避免碰撞的發生，進而增加自動化產線人員工作的安全性並減少人力監視的成本。
本論文以本實驗室自行開發之數位影像相關法為核心技術，應用於結構變形的精密量測、工業界的機械手臂快速運動分析以及人機共工的全場監測等領域，並將數位影像相關法整合低成本的硬體設備或者開源軟體，以降低成本及拓展應用範圍，開發符合工業價格及精度的量測系統。

關鍵字

數位影像相關法 ； 精密變形量測 ； 即時三維量測 ； 機械手臂 ； 動態物件取放 ； 人機共工

並列摘要

Digital image correlation (DIC) is a widely used optical measurement technique with non-contact and full-field advantages. Based on optimal image tracking algorithms, it provides sub-pixel accuracy for the displacement. Furthermore, velocity, acceleration, and strain can be determined by numerical differential of displacement. In this study, DIC is used to investigate the deformation of the cantilever plate subjected to applied force and thermal loading. The force deformation obtained by DIC is compared to the laser displacement sensor and finite element method to confirm the reliability of results. In the thermal deformation measurement, the thermal expansion coefficient of material is used to verify the results of DIC.
Regarding the industrial application, DIC is employed to examine the dynamic characteristics of the robotic arm under development. DIC combines with a high-speed camera is performed to measure the displacement, vibration, speed, and acceleration of robotic arm in high-speed motion, in order to evaluate potential problems of the developed robotic arm. The characteristic images are affixed to the robotic arm enable DIC to track its trajectory and joint angles and further reconstruct a robotic arm in virtual reality. Our system integrates OpenPose, image invariant moments, and pattern matching to monitor the full workspace of the robotic arm. This metrology system exemplifies a superior system compatibility. For instance, it has completed the basic architectural integration with real-time optical projection system, human motion prediction system, and obstacle avoidance path planning system. It utilizes merely the images to determine the shortest distance between the robotic arm and surrounding objects. When the distance is below the safety threshold, the worker will receive alarms to assist them to avoid risky circumstances. As a consequence, our system demonstrates outstanding abilities to raise the safety of the workspace and reducing personnel expenses.
This study uses DIC measurement system developed in our laboratory as the core technique to accomplish various applications, such as, monitoring of human-robot collaboration, robotic arm high-speed motion analysis, and precise measurement on structural deformation. Integrate DIC with low-cost hardware and open source software to reduce equipment costs and expand the industrial applicability. This study presents a simple and low-cost metrology system that satisfies the cost and accuracy criteria for industry.

並列關鍵字

Digital image correlation ； Precision deformation measurement ； Real-time three-dimensional measurement ； Robotic arm ； Dynamic pick-and-place ； Human-robot collaboration

參考文獻

[1] M. M. Frocht, Photoelasticity, vol. 1. J. Wiley, 1941.

[2] A. J. Durelli and V. J. Parks, Moiré analysis of strain. Englewood Cliffs, Prentice-Hall, 1970.

[3] T. Kreis, Handbook of holographic interferometry: optical and digital methods. Wiley, 2005.

[4] W. Yu, B. E. Fritz, N. Pernalete, M. Jurczyk, and R. V. Dubey, “Sensors assisted telemanipulation for maximizing manipulation capabilities of persons with disabilities.” 11th IEEE Symposium on Haptic Interfaces for Virtual Environment and Teleoperator Systems, Los Angeles, CA, USA, 2003.

[5] J. Mišeikis, K. Glette, O. J. Elle, and J. Torresen, “Multi 3D camera mapping for predictive and reflexive robot manipulator trajectory estimation.” IEEE Symposium Series on Computational Intelligence (SSCI), Athens, Greece, 2016.