Convolutional Neural Networks for Robust Classification of Drones

Dale, Holly; Jahangir, Mohammed; Baker, C.J.; Antoniou, Michail; Harman, Stephen; Ahmad, Bashar I.

doi:10.1109/radarconf2248738.2022.9764172

Cited by 12 publications

(6 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Previous work has reported the impact of SNR on classification performance [11]. Data from the UoB testbed are progressing the work on developing robust classifiers for challenging realistic conditions and an example of this is detailed in Table 3 from recently published results using Convolutional Neural Networks (CNN) for classifying drones and birds [12].…”

Section: Processing Capabilitiesmentioning

confidence: 99%

“…The vertical axis is radial velocity in m/s and the colour scale is signal power in dB.T A B L E 3 Percentage accuracies recorded with CNN trained and tested on drone and bird spectrograms taken from rural and urban locations. In total analysis was performed on 7623 spectrograms of drones and birds with a 60:30:10 split between training, test and validation phases[12]. Spectrogram of I3-D drone from (a) real data and (b) synthetic data generated using truth data from the real flight.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Development of a networked photonic‐enabled staring radar testbed for urban surveillance

Jahangir,

Griffiths,

White

et al. 2023

IET Radar Sonar & Navi

Self Cite

View full text Add to dashboard Cite

Urban surveillance of slow‐moving small targets such as drones and birds in low to medium airspace using radar presents significant challenges. Detecting, locating and identifying such low observable targets in strong clutter requires both innovation in radar hardware design and optimisation of processing algorithms. To this end, the University of Birmingham (UoB) has set‐up a testbed of two L‐band staring radars to support performance benchmarking using datasets of target and clutter from realistic urban environment. This testbed is also providing the vehicle to understand how novel radar architectures can enhance radar capabilities. Some of the challenges in installing the radar at the UoB campus are highligted. Detailed benchmarking results are provided from urban monostatic and bistatic field trials that form the basis for performance comparison against future hardware modification. The solution to the challenge of interfacing the radar to the external oscillators is described and stand‐alone bench tests with the candidate oscillators are reported. The testbed provides a valuable capability to undertake detailed analysis of performance of Quantum photonic‐enabled radar and allows for its comparison with conventional oscillator technology for surveillance of low observable targets in the presence of urban clutter.

show abstract

Section: Processing Capabilitiesmentioning

confidence: 99%

mentioning

confidence: 99%

Development of a networked photonic‐enabled staring radar testbed for urban surveillance

Jahangir,

Griffiths,

White

et al. 2023

IET Radar Sonar & Navi

Self Cite

View full text Add to dashboard Cite

show abstract

“…Drones are known to produce micro-Doppler shifts in radar echoes due to the modulation of incident radar waves by their rotating blades [13]. These shifts are unique radar signatures that can aid in the detection and classification of drones from other clutter, such as birds and humans [14][15][16][17][18][19][20]. Drones can be classified into three types based on the relative direction between the rotating plane of blades and the ground plane.…”

Section: Introductionmentioning

confidence: 99%

Exploring Radar Micro-Doppler Signatures for Recognition of Drone Types

Yan

Hu²,

Gong

et al. 2023

Drones

View full text Add to dashboard Cite

In this study, we examine the use of micro-Doppler signals produced by different blades (i.e., puller and lifting blades) to aid in radar-based target recognition of small drones. We categorize small drones into three types based on their blade types: fixed-wing drones with only puller blades, multi-rotor drones with only lifting blades, and hybrid vertical take-off and landing (VTOL) fixed-wing drones with both lifting and puller blades. We quantify the radar signatures of the three drones using statistical measures, such as signal-to-noise ratio (SNR), signal-to-clutter ratio (SCR), Doppler speed, Doppler frequency difference (DFD), and Doppler magnitude ratio (DMR). Our findings show that the micro-Doppler signals of lifting blades in all three drone types were stronger than those of puller blades. Specifically, the DFD and DMR values of pusher blades were below 100 Hz and 0.3, respectively, which were much smaller than the 200 Hz and 0.8 values for lifting blades. The micro-Doppler signals of the puller blades were weaker and more stable than those of the lifting blades. Our study demonstrates the potential of using micro-Doppler signatures modulated by different blades for improving drone detection and the identification of drone types by drone detection radar.

show abstract

“…This is by providing the means to consistently track the presence of any micro-Doppler components in the radar signature of drones (see Section 1.2). Micro-Doppler, which is a key target classification cue [20][21][22][23][24][25][26][27][28][29][30][31][32], is represented by spectral lines in Doppler spectrograms. They have a harmonic structure and originate from the motion of rotors on-board the UAS.…”

mentioning

confidence: 99%

“…For ATR, we use the machine-learning-based Multi-stage Decision Tree (MDT) in [28] that utilises kinematic and micro-Doppler features to discriminate between drone and non-drone targets. When tested on real representative data, it has a reasonable ATR accuracy compared with more advanced micro-Doppler classifiers, for example, neural-networks-based ones [21][22][23][24][25][26][27]. Below and unlike [32], we retrain the MDT model on MTT with PoT data to better illustrate how the consistent tracking of mUAS micro-Doppler components benefits ATR.…”

mentioning

confidence: 99%

A Bayesian track management scheme for improved multi‐target tracking and classification in drone surveillance radar

Ahmad,

Harman,

Godsill

2023

IET Radar Sonar & Navi

Self Cite

View full text Add to dashboard Cite

In this article, a simple, yet effective, Bayesian scheme for tracks maintenance, promotion, and deletion in drone surveillance radar is presented. It enables the simultaneous tracking of the target body and micro‐Doppler components that originate from the motion of rotors (if any) onboard an unmanned air system. This not only delivers more accurate multi‐target tracking, but also substantially improves the radar automatic target classification capability (e.g. discriminating between drone and non‐drone targets). Challenging and diverse real staring radar datasets are used here to demonstrate the efficacy and benefits of the proposed track management approach.

show abstract

Convolutional Neural Networks for Robust Classification of Drones

Cited by 12 publications

References 17 publications

Development of a networked photonic‐enabled staring radar testbed for urban surveillance

Development of a networked photonic‐enabled staring radar testbed for urban surveillance

Exploring Radar Micro-Doppler Signatures for Recognition of Drone Types

A Bayesian track management scheme for improved multi‐target tracking and classification in drone surveillance radar

Contact Info

Product

Resources

About