Robust estimation of chip clogging with supervised learning using tool surface image

The evaluation of chip clogging on the bottom surface of diamond-coated drills is traditionally performed through visual inspection by technicians. This study replicates this sensory evaluation by constructing a clogging discriminator using supervised machine learning with drill surface images. Image features ensuring high discrimination accuracy while minimizing wear or coating influence were identified. Verification results demonstrated that the histogram of oriented gradients and cumulative distribution function of luminance effectively construct a robust clogging discriminator, regardless of surface conditions. This method enables efficient clogging assessment, contributing to practical applications and improved drill performance evaluation in industrial settings.