Material Identification Using UWB

Abstract

Computer Vision can recognise the shape of an object but it is unable to see inside the object and determine the properties of the material. With the development of UWB radars it is possible to see under the skin of the observed object. In this paper, we propose using low cost commercial Ultra-Wide Band (UWB) modules for material identification that are, basically, designed for localization in wireless sensor networks. These modules are capable of measuring the RF impulse response of a channel (in this case the channel is the observed material) and, from the response, it is possible to classify materials using Machine Learning.