Autonomous process optimization of a tabletop CNC milling machine using computer vision and deep learning

CNC milling machines offer precision manufacturing across diverse materials but require off-machine quality checks and trial-and-error parameters selection. This study proposes a system that autonomously adjusts such parameters to improve surface quality and productivity. Composed of a deep learning-based monitoring apparatus capable of on-site surface roughness prediction with a 3.6 % mean error and a dataset of optimized parameters generated via multi-objective Bayesian optimization in only eleven attempts, it successfully conducted a fully autonomous trochoidal slotting operation, improving the final roughness by 36 %. The system, with further refinements, can be industrialized making autonomous machining accessible even to small and medium enterprises.