Since the advent of the first numerically controlled machine, the full automation of machine tools has been a central theme in manufacturing engineering research. Especially, the ability to change tools automatically due to tool wear or damage is an essential part of a machining system in the unmanned factory. Although a wide variety of tool failure sensing techniques have been developed recently, few have been successfully implemented in industries and show much limitation in the real application of on-line tool wear sensing.