Coherent, super-resolved radar beamforming using self-supervised learning

  title={Coherent, super-resolved radar beamforming using self-supervised learning},
  author={Itai Orr and Moshik Cohen and Harel Damari and Meir Halachmi and Mark Raifel and Zeev Zalevsky},
  journal={Science Robotics},
Description Self-supervised training of deep neural networks in unstructured environments enables enhanced angular resolution in radar systems. High-resolution automotive radar sensors are required to meet the high bar of autonomous vehicle needs and regulations. However, current radar systems are limited in their angular resolution, causing a technological gap. An industry and academic trend to improve angular resolution by increasing the number of physical channels also increases system… 
Self-Supervised Radio-Visual Representation Learning for 6G Sensing
In this work, a new label-free radio-visual co-learning scheme is formulated, whereby vision trains radio via cross-modal mutual information and the representation learnt by radiovisual self-supervision works well for a downstream sensing demonstrator, and outperforms its fully-supervised counterpart when less labelled data is used.
Physics and semantic informed multi-sensor calibration via optimization theory and self-supervised learning
A holistic approach to performing joint calibration of a camera-lidar-radar trio and proposes two targetless calibration methods within a cost minimization framework once via direct online optimization, and second via self-supervised learning (SSL).
Look, Radiate, and Learn: Self-supervised Localisation via Radio-Visual Correspondence
Results indicate that accurate radio target localisation can be automatically learned from paired radio-visual data without labels, which opens the door for vast data scalability and may prove key to realising the promise of robust radio sensing atop a unified perception-communication cellular infrastructure.


Cognitive radar antenna selection via deep learning
A convolutional neural network is constructed as a multi-class classification framework where each class designates a different subarray for antenna selection, thereby making antenna selection a cognitive operation.
Super Resolution Wide Aperture Automotive Radar
State-of-the-art automotive radars have an angular resolution of 1°, which is insufficient for estimating the objects shape and boundaries at long distance. In this paper, we design a novel
Vehicle Detection With Automotive Radar Using Deep Learning on Range-Azimuth-Doppler Tensors
This paper demonstrates a deep-learning-based vehicle detection solution which operates on the image-like tensor instead of the point cloud resulted by peak detection, and is the first to implement such a system.
Partially coherent radar unties range resolution from bandwidth limitations
A different type of ranging system is demonstrated, which possesses superior range resolution that is almost completely free of bandwidth limitations, and shows an improvement of two orders of magnitude, compared to standard coherent radars with the same bandwidth.
Enhanced Radar Imaging Using a Complex-Valued Convolutional Neural Network
This letter proposes a processing framework by which a complex-valued CNN (CV-CNN) is used to enhance radar imaging and shows the superiority of the proposed method on imaging quality and computational efficiency.
Deep Learning-based Object Classification on Automotive Radar Spectra
This work proposes to apply deep Convolutional Neural Networks (CNNs) directly to regions-of-interest (ROI) in the radar spectrum and thereby achieve an accurate classification of different objects in a scene.
Spectrum-based Single-Snapshot Super-Resolution Direction-of-Arrival Estimation using Deep Learning
A method for multi-target direction-of-arrival estimation for commercial FMCW radar systems in the automotive domain using an artificial neural network and indicates super-resolution like performance while significantly reducing and simultaneously limiting computation time compared to a maximum-likelihood search.
An Adversarial Super-Resolution Remedy for Radar Design Trade-offs
This work introduces a new method for circumventing radar design trade-offs and proposes the use of recent advances in computer vision, more specifically generative adversarial networks (GANs), to enhance low-resolution radar acquisitions into higher resolution counterparts while maintaining the advantages of the low- resolution parametrization.
Compressed Sensing based Single Snapshot DoA Estimation for Sparse MIMO Radar Arrays
In this paper a sparse antenna array with a low side lobe level is determined with a genetic algorithm and a cost function and an investigation is performed what difference in the radar cross section of two targets in the same range-Doppler cell can be achieved.
Spatial Compressive Sensing for MIMO Radar
The coherence and isotropy concepts are used to establish uniform and non- uniform recovery guarantees within the proposed spatial compressive sensing framework, and it is shown that non-uniform recovery is guaranteed if the product of the number of transmit and receive elements, MN, scales with K(logG)2, which is proportional to the array aperture and determines the angle resolution.