Vision-Based Fault Classification for Monitoring Industrial Robot

Abstract

In this paper, the sparse optical flow (SOF), in form of three-way array, is used as motion description. Based on unfolded SOF, we apply PCA to get the principal component (PC). Then, two statistics T <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> and SPE are defined for monitoring the motion process of the industrial robot. To improve the performance of the fault detection accuracy, we apply multi-manifold learning algorithm with the labeled root fault area frame detected by PCA method. The proposed method can detect the fault more effectively, while reducing the false alarm rate significantly. Experiment of robotic-arm based marking system (RABMS) is taken to evaluate the performance of the proposed method. The results demonstrate the capability of the proposed methods.