Search

This paper presents measurements of Radar Cross Section (RCS) of five Unmanned Aerial Vehicles (UAVs), comprising both consumer grade and professional small drones, collected in a semi-controlled environment as a function of azimuth aspect angle, polarization and frequency in the range 8.2-18 GHz. The experimental setup and the data pre-processing, which include coherent background subtraction and range gating procedures, are illustrated in detail. Furthermore, a thorough description of the calibration process, which is based on the substitution method, is discussed. Then, a first-order statistical analysis of the measured RCSs is provided by means of the Cramér-von Mises (CVM) distance and the Kolmogorov-Smirnov (KS) test. Finally, radar detection performance is assessed on both measured and bespoke simulated data (leveraging the results of the developed statistical analysis), including, as benchmark terms, the curves for non-fluctuating and Rayleigh fluctuating targets.

This paper deals with Radar Cross Section (RCS) measurements of five small Unmanned Aerial Vehicles (UAVs) in a semi-controlled environment as a function of azimuth aspect angle, polarization, and frequency in the range 8.2-18 GHz. The experimental setup and the data pre-processing, which include coherent background subtraction, range gating procedures, and calibration, are presented. Then, a thorough statistical analysis of the measured RCSs is provided by means of the Cram´er–von Mises (CVM) distance and the Kolmogorov–Smirnov (KS) test.

We introduce FocalPose, a neural render-and-compare method for jointly estimating the camera-object 6D pose and camera focal length given a single RGB input image depicting a known object. The contributions of this work are twofold. First, we derive a focal length update rule that extends an existing state-of-the-art render-and-compare 6D pose estimator to address the joint estimation task. Second, we investigate several different loss functions for jointly estimating the object pose and focal length. We find that a combination of direct focal length regression with a reprojection loss disentangling the contribution of translation, rotation, and focal length leads to improved results. We show results on three challenging benchmark datasets that depict known 3D models in uncontrolled settings. We demonstrate that our focal length and 6D pose estimates have lower error than the existing state-of-the-art methods., Accepted to CVPR2022. Code available at http://github.com/ponimatkin/focalpose

The goal of this work is to efficiently identify visually similar patterns in images, e.g. identifying an artwork detail copied between an engraving and an oil painting, or recognizing parts of a night-time photograph visible in its daytime counterpart. Lack of training data is a key challenge for this co-segmentation task. We present a simple yet surprisingly effective approach to overcome this difficulty: we generate synthetic training pairs by selecting segments in an image and copy-pasting them into another image. We then learn to predict the repeated region masks. We find that it is crucial to predict the correspondences as an auxiliary task and to use Poisson blending and style transfer on the training pairs to generalize on real data. We analyse results with two deep architectures relevant to our joint image analysis task: a transformer-based architecture and Sparse Nc-Net, a recent network designed to predict coarse correspondences using 4D convolutions. We show our approach provides clear improvements for artwork details retrieval on the Brueghel dataset and achieves competitive performance on two place recognition benchmarks, Tokyo247 and Pitts30K. We also demonstrate the potential of our approach for unsupervised image collection analysis by introducing a spectral graph clustering approach to object discovery and demonstrating it on the object discovery dataset of \cite{rubinstein2013unsupervised} and the Brueghel dataset. Our code and data are available at http://imagine.enpc.fr/~shenx/SegSwap/., Comment: add results of unsupervised saliency detection

This paper deals with the analysis of simultaneously collected co- and cross-polarized bistatic sea-clutter returns with special emphasis on their representation as a Spherically Invari-ant Random Process (SIRP). The study is conducted by using appropriate testing procedures involving the complex envelope of the measured data that provide both first- and higher-order compatibility conditions. The results highlight that the SIRP model is a good candidate for the representation of bistatic coherent clutter, and usually the coherence time of the SIRP texture is longer than that in the monostatic case.

In this paper, a novel multi-channel waveform agile radar testbed - ASTAP (Advanced Space-Time Adaptive Processing) with eight transmit channels and a single receive channel to form a co-located Multiple-Input Single-Output (MISO) radar is presented. Based on current configuration, the ASTAP radar system possesses agility for synthesizing and transmitting independent waveforms via each transmit channel simultaneously via multi-channel Arbitrary Waveform Generator (AWG). Hardware imperfections influence on digitally created waveforms including limited-bits DAC (Digital-to-Analog Converter) quantization, inter-channel time- and phase-skew (and/or jitter), non-linearities introduced by high-frequency devices (RF amplifiers and mixers etc) and antenna dispersion effects has been studied experimentally.An end-to-end system-level digital calibration using Over-the-Air (OTA) channel measurements to minimize system hardware imperfections prior to transmission is proposed. Calibration accuracy for waveform transmit ambiguity functions, orthogonality in MISO transmissions, received signals separation and beamforming process for the synthesis of target azimuth distributions is examined and quantified. Front-end performance with different waveforms types for colored transmission and simultaneous MIMO is demonstrated and analysed. Extraction of the signals corresponding to each transmit channel from the composite received signal in a single receive channel exploiting waveforms orthogonality is studied experimentally.

This paper deals with the effects of Power Amplifiers (PAs) on radar waveforms designed to ensure spectral compatibility with other Radio Frequency (RF) wireless systems cohabitating in the same frequency band. To this end, a specific hardware-in-the-loop test-bed composed of a Personal Computer (PC), an Arbitrary Waveform Generator (AWG), a nonlinear PA, and a Signal Analyser (SA) is designed. It allows to gather measurements of the amplified signal and to grasp unwanted modifications of its spectral features due to the PA nonlinear distortion. Besides, it permits a quantitative assessment of the discrepancies between the nominal and the actual interference power injected by the radar in the shared frequency bandwidths. To alleviate the spectral distortions induced by the PA and improve spectral coexistence, the use of a Digital Pre-Distortion (DPD) stage upstream the PA is investigated. Modeling techniques together with parameters inference are discussed and applied to the measured waveforms. The obtained experimental results highlight that pairing a suitable waveform design strategy with a DPD processing represents a viable means to mitigate the PA distortions and to realize spectral coexistence.

A new algorithm for 3D localization in multiplat-form radar networks, comprising one transmitter and multiple receivers, is proposed. To take advantage of the monostatic sensor radiation pattern features, ad-hoc constraints are imposed in the target localization process. Therefore, the localization problem is formulated as a non-convex constrained Least Square (LS) optimization problem which is globally solved in a quasi-closed-form leveraging Karush-Kuhn-Tucker (KKT) conditions. The results corroborate the effectiveness of the new strategy which is capable of ensuring a lower Root Mean Square Error (RMSE) than counterpart methodologies, especially in the low Signal to Noise Ratio (SNR) regime.

Unmanned Aerial Vehicles (UAVs) have become widely used in civilian and military applications. There is a recognised need for UAVs surveillance against their malicious use. In this context, radar has emerged as a powerful tool in detecting and tracking UAVs. However, the task is quite complicate due to the low-altitude, low velocity and low radar cross section of small UAVs. In this paper, a smart radar signal processing solution in the presence of multiple unresolved UAVs is devised and applied to UAVs surveillance, which fully capitalizes on the coherent integration, constant false alarm rate detection, target multiplicity test and sparse methods to achieve low computational complexity and high angular resolution. At the analysis stage, numerical simulations are carried out to highlight the capability of proposed solution to detect UAVs and resolve them when they are closely spaced.

Local perception navigation methods allow agricultural robots to accurately track crop row structures while performing automated farming tasks. The integration of these methods as a part of a fully autonomous navigation solution requires continuous assessment of their reliability since they rely solely on sensor data in a changing and unpredictable environment. This paper presents a data-driven monitoring approach for the task of structure-based navigation in agriculture. The proposed method applies semi-supervised anomaly detection, aiming to learn a model of normal scene geometry that characterizes a domain of reliable execution of the considered task. To this end, a convolutional neural network was trained in one-class classification fashion on Hough representations of LiDAR point clouds. In experimentation, the learned normal model was used to derive a confidence measure for a LiDAR-based tracking algorithm allowing its integration as a part of a hybrid navigation solution in vineyards for a commercial robotic platform.

Considering a SpaceBorne Radar (SBR) for Space Situational Awareness (SSA), the Fading Occurrence Probability (FOP) is computed for fluctuating targets in weak plasma turbulence as a function of the scintillation index $S_{4}$ . While previous works considered a Rayleigh model for both target fluctuation and plasma scintillation, this paper extends the FOP via the Cumulative Distribution Function (CDF) of the echo power modulated by either non-fluctuating, Rayleigh, and Rice target fluctuations along with Rice plasma scintillation. Downstream physical insights on both channel and target phenomenology leading to such statistical models, numerical results corroborated by Monte Carlo (MC) simulations show that a larger value of the scintillation index $S_{4}$ increases the FOP on echoes. In addition, the aforementioned simulation curves provide a conceptual framework to empirically estimate the scintillation index $S_{4}$ in a monostatic radar configuration.

This paper deals with the use of Reconfigurable Intelligent Surfaces (RISs) for radar surveillance in Non-Line Of Sight (N-LOS) scenarios. First of all, the geometry of the scene and the new system concept is described with emphasis on the required operative modes and the role played by the RIS. Then, the specific radar equation (including the RIS effect) is developed to manage the coverage requirements in the challenging region where the LOS is not present. Hence, a digression on the use of the radar timeline for the new operative mode is presented together with a discussion on the resolution issues for the range, azimuth, and Doppler domains. Finally, the interplay among the system parameters and, in particular, those involving the RIS is discussed and analyzed via numerical simulations.

The problem of missing sensor measurements can emerge in a variety of radar signal processing applications as for instance beamforming, direction of arrival estimation, interference cancellation, and target detection. The mentioned applications rely on reliable data covariance matrix estimates and suitable procedures, based on the expectation-maximization (EM) algorithm, have been proposed in the open literature to cope with the lack of some entries within specific spatial snapshots. In this paper, the effectiveness of a recent structured covariance matrix estimator [1], accounting for missing data and leveraging possible structural knowledge, is assessed on measured data. Specifically, the estimation procedure is framed in the context of two practically relevant radar applications: beamforming and detection of the number of sources. At the analysis stage, results highlight the effectiveness of the procedure to tackle missing data in the considered radar scenarios.