Intelligent Material Classification and Identification Using a Broadband Millimeter-Wave Frequency Comb Receiver

Abstract

Millimeter-wave radars offer a practical solution to distinguish objects made of different materials, shapes, and compositions. In this work, radar classification of various materials is demonstrated using a broadband millimeter-wave CMOS integrated receiver. The receiver is used to record the transmitted power through multiple solid materials at various distances from the receiver in the W-band (75–110 GHz). Three supervised machine learning tools are trained by the recorded data to classify these materials into different categories. The trained classifiers were used to predict material and thickness of objects with varying distances from the receiver with accuracy levels of higher than 96% in material classification and 88% in thickness classification.