Condition Monitoring of Integrated Robots Based on Fiber Optic Sensing and Deep Learning

Abstract：Intelligent robots are being increasingly used in modern manufacturing. Condition monitoring is one of the crucial technologies for their safe operations. At the current stage, due to the drawbacks of electrical sensors such as poor embeddability and poor electromagnetic resistance, condition monitoring systems applied to integrated industrial robots are relatively limited. In response to this demand, we are developing a condition monitoring system based on fiber optic sensing and deep learning for the integrated industrial robot. Since fiber optic sensors have the characteristics of anti-electromagnetic interference, small size, lightweight, excellent flexibility, great embeddability, and will not produce electrical signals that interfere with or damage the electrical circuits of industrial robots, they can be ideally embedded into integrated industrial robots to monitor their operation status more effectively and directly. This report will focus on how to integrate fiber optic sensors into an integrated industrial robot and introduce new approaches based on deep learning for anomaly detection and collision identification of the integrated industrial robot. We believe that the proposed condition monitoring system holds significant potential for application in intelligent industrial robots and provides robust support for the ongoing advancement of Industry 4.0.