Advanced classification techniques for drone payloads

Abstract

This chapter presents a summary of radar-based classification approaches developed for small drones carrying payloads. Specific focus is given to three types oftechniques that were validated on the same multistatic radar data set collected usingthe University College London (UCL)-netted radar NetRAD. These techniquesused, respectively, features extracted from the centre of mass and bandwidth of themicro-Doppler signatures; different radar data domains generated from the micro-Doppler data to be processed by pretrained Convolutional Neural Networks(CNNs) and spectral kurtosis analysis on the micro-Doppler.