In-Wall Object Recognition based on SAR-like Imaging by UWB-Radar

Inhalt:
In this paper an approach for object recognition by backscattered Ultra Wideband (UWB) signals on the basis of a reference data set gained by a hybrid ray tracing/FDTD method is presented. This approach is part of a UWB imaging and localization concept for emergency scenarios. Here a security robot is moving at a certain track through a room. The security robot has the task of providing a rough image and has to inspect the objects in the room. The proposed recognition method can also be used to assess the intactness of the building construction elements. Usually such elements are composed of different materials, often including a metallic core. The UWB frequency range selected ensures a sufficient penetration of the material as well as a resolution of about 3mm for the outer objects surface and exhibits a slightly reduced resolution for an enclosed object. The biggest impact is here the possibility to completely move around the object. The imaging and the subsequent object recognition is performed in several steps which start with an imaging of the outer object shape using an IBST algorithm which evaluates the wavefronts and continues with an estimation of the objects permittivity. A modified IBST algorithm then determines the enclosed objects form