A Feature-Based Approach for Loaded/Unloaded Drones Classification Exploiting micro-Doppler Signatures

Chebychev moments based Drone Classification, Recognition and Fingerprinting

“…A Ku band dataset, containing returns from 10 (1 bird, 3 fixed wing and 6 multi rotor) types of targets and acquired by Plextek DTS, is used for the Drone vs Bird and Fixed Wing vs Multi Rotor tasks, while the dataset collected by the University College London comprising loaded and unloaded drones using the S-band radar system NetRAD [14] is used to assess the capability to discriminate between loaded/unloaded drones and discriminating the payload weight. Details of the Ku band dataset are reported in Table 1, while in [6,14] can be found the details of the NetRAD data. To evaluate the effective- ness of the classification procedure proposed in Section 2, the considered figure of merit for the Ku band dataset are the accuracy and the average F1-score, which results to be more reliable when dealing with unbalanced datasets, while accuracy only is considered for the S-band data.…”

“…In [5], the first investigation on the effect of different payloads was performed, proposing additional strategies to classify the radar returns in a multi-static radar system. While in [6] the payload recognition is addressed by using a spectral kurtosis based approach, interestingly other payload types have been also investigated in the literature recently such as dynamic payloads in [7]. As a matter of fact, micro-Doppler, that can be seen as additional frequency modulations induced by small displacement, rotation or vibration of secondary parts of the object under observation, have been widely exploited for target classification and micro-motion analysis, including hand-gesture, over the last decade [8][9][10][11].…”

Multi-Frequency Radar Micro-Doppler Based Classification of Micro-Drone Payload Weight

“…A comparatively more limited body of literature is published on the topic of micro-Doppler based classification of unloaded and loaded drones. A single data-set collected by University College London's (UCL) multistatic S-Band radar system, NETRAD (Doughty, 2008), has been used in a number of publications (Fioranelli et al, 2015;Ritchie et al, 2017;Patel et al, 2019;Pallotta et al, 2020), to test the accuracy of a variety of payload classification algorithms. In this data-set, a DJI Phantom 2 was loaded with a variety of payloads between 0 and 600 g and made to either hover or fly towards the central radar node.…”

“…In contrast, for payload classification of the drone in flight, a maximum classification accuracy of 95% was achieved when using singular value decomposition (SVD) derived features. A similar feature based approach was used on the same data-set in a more recent publication by Pallotta et al (2020), differing mainly through the use of spectral kurtosis (Dwyer, 1983) to extract features from the spectrograms. The algorithm produced by Pallotta et al (2020) achieved classification accuracies for the hovering drone of 92.61%, marginally lower than the result reported by Ritchie et al (2017).…”

Classification of micro-Doppler radar hand-gesture signatures by means of Chebyshev moments

“…In this respect, an effective tool to properly perform the gesture sensing and recognition is the well-known micro-Doppler effect [3]. As a matter of fact, micro-Doppler, that can be seen as additional frequency modulations induced by small displacement, rotation or vibration of secondary parts of the object under observation, have been widely exploited for target classification and micro-motion analysis, including hand-gesture, over the last decade [4]- [7].…”