Target tracking in industrial multi-sensor short-range radar applications using Doppler and amplitude information

This paper treats localization and tracking of small extended targets in short-range applications with a multi-sensor Doppler radar network. It particularly addresses cases where the point target assumption is not valid. Hence, the target extensions and the fluctuating radar cross section (RCS) are considered during the raw data processing and in the tracking algorithm. Here, a metallic cuboid with a priori known alignment and backscattering properties is considered. Such target shapes often appear as raw workpieces in industrial milling and drilling applications. A multi-sensor platform for Doppler-only localization of the extended target in a bounded environment is introduced. An approach to combine additional amplitude information (AI) observed by the sensors is presented. This approach is available in the considered application, where closely-spaced sensors in their specific sensor-target geometry provide information about the alignment from amplitude evaluation. The coordinate estimates are then utilized in the tracker initialization procedure to increase the tracking reliability. Results from electromagnetic RCS simulations are given and the estimation algorithm is presented, which delivers reliable coordinate estimates from the observed AI by probability mixing. Simulation results of the initialization procedure and obtained improvement for tracking a single extended cube are shown.