It remains a challenge to acoustically localise drones, especially when there are multiple sound sources in the environment. The aim of this paper is to propose and validate an approach for joint acoustic localisation of drones and other background sources using microphone array measurements. This approach consists of a frequency-domain beamforming step, where a discrete number of frequencies in the range of 140 to 1200 Hz are beamformed individually. The location of the maximum output is identified and plotted in the azimuth and elevation direction per frequency and time step. The next step is to extract and separate the trajectories from this plot. The validation shows the correctness of this approach, together with the remaining challenges. Although sources are extracted and separated correctly, overlapping sources are sometimes considered as only one source. Moreover, in low-level drone noise or high-level background noise measurements, erroneous sound sources are identified. Nevertheless, the results show that drones can be localised over long distances, up to 712 m and 593 m for a quadcopter drone in azimuth and elevation direction respectively, and 317 m and 305 m for a fixed-wing drone in azimuth and elevation direction respectively.

Abstract: Detecting malicious drones using aerial surveillance cameras is challenging when the distance is large, because the drone then occupies only a few pixels. Current optical detection methods rely mostly on visual appearance features. Hence, they struggle to differentiate drones from other flying objects, especially birds, when the apparent object size is small. Fortunately, the observed trajectory over time can help improve the differentiation accuracy. Here, we propose to combine classification neural networks of the object’s trajectory features and visual appearance features. We train and test the networks using our dataset containing infrared videos of drones and birds, where the variation of drone configurations and flight patterns is relatively larger than other publicly available datasets. We show that, particularly for small objects with high motion, the inclusion of trajectory features for visual classification achieves up to 22% higher frame-wise classification accuracy compared to when only visual appearance features are used. We further demonstrate that integrating both feature types provides improved accuracy over all of the considered trajectories, with 4% more of the trajectories being classified correctly. Consistent results are also shown on an open dataset, confirming the generalizability. Our study demonstrates the crucial role of information beyond the frame-wise visual appearance features in extending the operational range of aerial surveillance cameras.

This study covers three aspects of acoustic localisation of drones using a microphone array. First, it assesses a grid-free approach, using differential evolution, to estimate the three-dimensional position of a drone. It is found that this is indeed possible for the drone in the near-field. For larger distances, it still provides the angular position of the drone. Second, the study emphasizes the essence of localisation over small frequency bands with the bands jointly spanning a large frequency range to reveal the presence of multiple sound sources and maximise the drone localisation range. Third, it addresses the localisation ranges for six different drones.

Correction Notice • Place s5 in front of the symbols ‘greater than or equal to (≥)’ and ‘less than (<)’ in Equation 2, as shown below. (Formula Presented). • Replace the correlation coefficient ‘0.886’ showed in Table 6 to ‘-0.886’. This correction also apply to the paragraph below Section 3 (Correlation between Annoyance and Drone Characteristics) and in the last paragraph of the Conclusions. This negative value indicates an inverse relationship between the installation ratio (d/D) and the annoyance computed using the More PA model. • Replace ‘0.337 (Zwicker PA model)’ by ‘0.362 (Willemsen PA model)’ in the paragraph below Section 3 (Correlation between Annoyance and Drone Characteristics). This value (0.362) is also reported in Table 6.

Due to technological advances in the drone industry, security threats induced by unmanned aerial vehicles (UAVs) are becoming more relevant. Fast and accurate localisation systems need to be designed. One approach is localisation of UAVs by their sound using acoustic techniques. So far, a systematic performance assessment of acoustic techniques for drone localisation, based on real-world data, is lacking. This work presents a comparison of selected techniques using real-world measurement data. The achieved performance serves as a baseline for future design of novel localisation methods. Three techniques are chosen. The first technique estimates the time-difference-of-arrival (TDOA) using generalised cross-correlation with phase transform weighting (GCC-PHAT). The second technique is differential evolution, which approaches the localisation task as a global optimisation problem. The third technique is conventional frequency domain beamforming. Real-world data of 5 quadrotor UAVs were used acquired with an acoustic microphone-array. The performance of the techniques is assessed using the absolute error between the estimated source location and the true source location obtained from the onboard GPS tracker of the drones. GCC-PHAT and differential evolution attempt to estimate the drone position in one or few steps. They have a much shorter runtime than beamforming, which is an exhaustive grid search algorithm. However, these techniques result in lower detection ranges and accuracy compared to beamforming.

Threats posed by drones urge defence sectors worldwide to develop drone detection systems. Visible-light and infrared cameras complement other sensors in detecting and identifying drones. Application of Convolutional Neural Networks (CNNs), such as the You Only Look Once (YOLO) algorithm, are known to help detect drones in video footage captured by the cameras quickly, and to robustly differentiate drones from other flying objects such as birds, thus avoiding false positives. However, using still video frames for training the CNN may lead to low drone-background contrast when it is flying in front of clutter, and omission of useful temporal data such as the flight trajectory. This deteriorates the drone detection performance, especially when the distance to the target increases. This work proposes to pre-process the video frames using a Bio-Inspired Vision (BIV) model of insects, and to concatenate the pre-processed video frame with the still frame as input for the CNN. The BIV model uses information from preceding frames to enhance the moving target-to-background contrast and embody the target’s recent trajectory in the input frames. An open benchmark dataset containing infrared videos of small drones (< 25 kg) and other flying objects is used to train and test the proposed methodology. Results show that, at a high sensor-to-target distance, the YOLO algorithms trained on BIV-processed frames and concatenation of the BIV-processed frames with still frames increase the Average Precision (AP) to 0.92 and 0.88, respectively, compared to 0.83 when it is trained on still frames alone.