Machine learning for in-process end-point detection in robot-assisted polishing using multiple sensor monitoring

The decision on polishing operation stopping time when employing a robot-assisted polishing machine is a critical issue for the full automation of the polishing process. In this paper, a machining learning approach based on artificial neural networks was developed using multiple sensor monitoring data to realize an intelligent system capable to determine the state of the polishing process in terms of target surface roughness achievement. During the experimental tests, surface roughness measurements were performed on each polished workpiece and the acquired sensor signals were analyzed and processed by applying two kinds of feature extraction procedures: statistical features extraction and principal component analysis. By feeding diverse types of feature pattern vectors to artificial neural networks, a highly accurate classification of the polishing process state was obtained using the principal component feature pattern vectors.