High-resolution Radar Sensors Research Articles

Efficient Target Classification Based on Vehicle Volume Estimation in High-Resolution Radar Systems

In this paper, we propose a method for efficient target classification based on the spatial features of the point cloud generated by using a high-resolution radar sensor. The frequency-modulated continuous wave radar sensor can estimate the distance and velocity of a target. In addition, the azimuth and elevation angle of the target can be estimated by using a multiple-input and multiple-output antenna system. Using the estimated distance, velocity, and angle, the 3D point cloud of target can be generated. From the generated point cloud, we extract the point cloud for each individual target using the density-based spatial clustering of application with noise method and a camera mounted on the radar sensor. Then, we define the convex hull boundaries that enclose these point clouds in both 3D and 2D spaces obtained by orthogonally projecting onto the xy, yz, and zx planes. Using the vertices of convex hull, we calculate the volume of the targets and the areas in 2D spaces. Several feature points, including the calculated spatial information, are numerized and configured into feature vectors. We design an uncomplicated deep neural network classifier based on minimal input information to achieve fast and efficient classification performance. As a result, the proposed method achieved an average accuracy of 97.1%, and the time required for training was reduced compared to the method using only point cloud data and the convolutional neural network-based method.

Sparsity-Robust Feature Fusion for Vulnerable Road-User Detection with 4D Radar

Detecting vulnerable road users is a major challenge for autonomous vehicles due to their small size. Various sensor modalities have been investigated, including mono or stereo cameras and 3D LiDAR sensors, which are limited by environmental conditions and hardware costs. Radar sensors are a low-cost and robust option, with high-resolution 4D radar sensors being suitable for advanced detection tasks. However, they involve challenges such as few and irregularly distributed measurement points and disturbing artifacts. Learning-based approaches utilizing pillar-based networks show potential in overcoming these challenges. However, the severe sparsity of radar data makes detecting small objects with only a few points difficult. We extend a pillar network with our novel Sparsity-Robust Feature Fusion (SRFF) neck, which combines high- and low-level multi-resolution features through a lightweight attention mechanism. While low-level features aid in better localization, high-level features allow for better classification. As sparse input data are propagated through a network, the increasing effective receptive field leads to feature maps of different sparsities. The combination of features with different sparsities improves the robustness of the network for classes with few points.

RF Glass Technology Is Going Mainstream: Review and Future Applications

Driven by the increasing demand for high-throughput communication links and high-resolution radar sensors, the development of future wireless systems pushes at ever greater operating frequencies. By analogy, high-performance computing (HPC) systems with high-bandwidth I/Os have become a mainstream solution to address multi Gbit/s data rates. Heterogeneous integration technologies play a vital role here in enhancing the performance and functional density, along with reducing the size and costs of such RF systems. In line with this trend, many passive components, which have once been monolithically integrated, are now implemented on ceramic or polymer-based packages. Besides these long-established material and process technologies, considerable efforts have also recently been devoted to the development of RF components on glass and glass-ceramics. Spurred by their low dielectric losses, extremely smooth surfaces, and excellent dimensional stability, glass technologies are emerging as a promising alternative for high-volume and high-performance RF applications. In this article, relevant glass materials and key enabling technologies are reviewed and put into context with well-established RF substrate technologies. Another focus is set on the latest glass-based packaging and interposer solutions ranging from MHz-to-THz frequencies. To showcase the development activities and practical accomplishments of the RF glass technology, also a large variety of key components is presented. Finally, the paper concludes by discussing future research and development directions of RF glass devices.

Multi‐view convolutional neural network‐based target classification in high‐resolution automotive radar sensor

In this study, a target classification method based on point cloud data in a high-resolution radar sensor is proposed. By using multiple antenna elements arranged in horizontal and vertical directions, pedestrians, cyclists and vehicles can be expressed as point cloud data in the three-dimensional (3D) space. To perform target classification using the spatial characteristics (i.e. length, height and width) of the target, the 3D point cloud data is orthogonally projected onto the xy, yz and zx planes, respectively, and three types of images are generated. Then, a multi-view convolutional neural network (CNN)-based target classifier using those three images as inputs is designed. To this end, a method for synthesising the detection results of three directions in series or in parallel is proposed. The proposed classifier can learn the spatial characteristics of the target by using the detection results of multiple viewpoints. Compared to the CNN-based classifier that uses only the detection result of a single plane as input, the proposed method shows 4.5%p higher classification accuracy in terms of the target with the lowest classification accuracy. In addition, the proposed multi-view CNN structure shows improved classification performance and shorter training time compared to the well-known deep learning methods for image classification.

Phase Noise Monitoring in Cascaded Systems for High-Resolution Automotive Radar Sensors

The trend toward a higher degree of automation in modern cars demands radar sensors providing a significantly increased angular resolution. This is achieved by increasing the number of transmit and receive channels, which is most commonly done by cascading several radar monolithic microwave integrated circuits (MMICs). As for single MMIC radar solutions, the phase noise (PN) contained in the transmit signal is one of the most dominant signal distortions limiting detection sensitivity and range also in cascaded radar systems. In this work, we present a novel method to estimate the mean PN power spectral density (PSD) of two radar MMICs in operation. This estimation result is used to detect whether the frequency-generating circuit of the involved MMICs generate PN out of specification. Different to most of the existing methods, the proposed approach is based on an application relevant frequency modulated continuous wave (FMCW) signal, which is actually used to perform a radar measurement, instead of a continuous wave (CW) signal. Finally, the presented concept is verified based on simulations and with the help of measurement data.

Radar Odometry on $SE(3)$ With Constant Velocity Motion Prior

This letter presents an approach to estimating vehicle odometry on SE(3) using a high resolution automotive radar sensor. We combine a constant velocity SE(3) motion prior with a 3D radar point cloud measurement model in a sliding window optimization scheme. We leverage radar's highly precise radial velocity measurement to compensate for point cloud sparsity and improve data association. Our approach is tested using real-world measurements from a prototype high-resolution radar sensor. We demonstrate the results of this approach for 6D motion estimation. In addition, we show that our approach is comparable to state-of-the-art SE(2) radar odometry estimates while running a full order of magnitude faster.

Constraint-Based Hierarchical Cluster Selection in Automotive Radar Data.

High-resolution automotive radar sensors play an increasing role in detection, classification and tracking of moving objects in traffic scenes. Clustering is frequently used to group detection points in this context. However, this is a particularly challenging task due to variations in number and density of available data points across different scans. Modified versions of the density-based clustering method DBSCAN have mostly been used so far, while hierarchical approaches are rarely considered. In this article, we explore the applicability of HDBSCAN, a hierarchical DBSCAN variant, for clustering radar measurements. To improve results achieved by its unsupervised version, we propose the use of cluster-level constraints based on aggregated background information from cluster candidates. Further, we propose the application of a distance threshold to avoid selection of small clusters at low hierarchy levels. Based on exemplary traffic scenes from nuScenes, a publicly available autonomous driving data set, we test our constraint-based approach along with other methods, including label-based semi-supervised HDBSCAN. Our experiments demonstrate that cluster-level constraints help to adjust HDBSCAN to the given application context and can therefore achieve considerably better results than the unsupervised method. However, the approach requires carefully selected constraint criteria that can be difficult to choose in constantly changing environments.

Early Gesture Recognition With Reliable Accuracy Based on High-Resolution IoT Radar Sensors

Early recognition of gestures gives a better user experience during human–computer interaction in a near real-time way. However, the inadequate information accumulated in the early stages of the gesture will make the machine ambiguous and respond wrongly. Therefore, the challenge of early recognition lies in choosing the right time to trigger gesture prediction and using limited information to achieve reliable accuracy. In this article, an early gesture recognition method is proposed to achieve almost the same accuracy as a complete gesture sequence. For this purpose, we design a new architecture to modify the current feature sequence according to the existing information and introduce a new loss function to maximize the probability of correct gesture label as early as possible. At the same time, gesture recognition is triggered when the maximum probability of model output is greater than a set threshold. The publicly available data set used for evaluation comes from Google’s Project Soli sensor. Experimental results show that the proposed method for early recognition can achieve a gesture recognition rate of 88.85% with an average time saving of 69.47%. Compared with the results of all sequence recognition, the recognition rate drops by only 1.7%. In addition, the proposed method based on the data set constructed by terahertz radar also shows similar early recognition performance and can be applied to other radar sensors.

Extended Object Tracking Using Spatially Resolved Micro-Doppler Signatures

Recent progress in radar development enable high-resolution radar sensors to resolve multiple detections for an extended object. One of the main challenges in radar-based object detection and tracking is the handling of an uncertain measurement model problem dealing with wandering dominant scatter points on the extended object surface depending on the relative orientation and motion. This phenomenon needs to be considered for near extended object tracking in critical driving situations, e.g., for safety applications, which require fast and accurate target state estimation in high dynamic scenarios. This paper presents a method to detect rotating wheels of approaching targets. Using the position and velocity of detected wheels as input for a tracking procedure, instead of target detections, relaxes the uncertain measurement model problem. A generic radar signal processing method to detect the position and bulk velocities of rotating wheels for approaching target vehicles is presented. The method evaluates the micro-Doppler signals, generated by rotating wheels, of close targets. These wheel hypotheses serve as input for a subsequent sequential Monte Carlo tracking framework. The presented methods are evaluated on real experimental data where a target vehicle is used to reconstruct safety-critical dynamic evading maneuvers.

Tracking Multiple Vehicles Using a Variational Radar Model

High-resolution radar sensors are able to resolve multiple detections per object and, therefore, provide valuable information for vehicle environment perception. For instance, multiple detections allow us to infer the size of an object or to measure the object’s motion more precisely. Yet, the increased amount of data raises the demands on tracking modules; measurement models that are able to process multiple detections for an object are necessary and measurement-to-object associations become more complex. This paper presents a new variational radar model for tracking vehicles using radar detections and demonstrates how this model can be incorporated into a random-finite-set-based multi-object filter. The measurement model is learned from actual data using variational Gaussian mixtures and avoids excessive manual engineering. In combination with the multi-object tracker, the entire process chain from raw measurements to the resulting tracks is formulated probabilistically. The presented approach is evaluated on experimental data, and it is demonstrated that the data-driven measurement model outperforms a manually designed model.

Computational Burden Resulting from Image Recognition of High Resolution Radar Sensors

This paper presents a methodology for high resolution radar image generation and automatic target recognition emphasizing the computational cost involved in the process. In order to obtain focused inverse synthetic aperture radar (ISAR) images certain signal processing algorithms must be applied to the information sensed by the radar. From actual data collected by radar the stages and algorithms needed to obtain ISAR images are revised, including high resolution range profile generation, motion compensation and ISAR formation. Target recognition is achieved by comparing the generated set of actual ISAR images with a database of ISAR images generated by electromagnetic software. High resolution radar image generation and target recognition processes are burdensome and time consuming, so to determine the most suitable implementation platform the analysis of the computational complexity is of great interest. To this end and since target identification must be completed in real time, computational burden of both processes the generation and comparison with a database is explained separately. Conclusions are drawn about implementation platforms and calculation efficiency in order to reduce time consumption in a possible future implementation.

Computational Burden Resulting from Image Recognition of High Resolution Radar Sensors

This paper presents a methodology for high resolution radar image generation and automatic target recognition emphasizing the computational cost involved in the process. In order to obtain focused inverse synthetic aperture radar (ISAR) images certain signal processing algorithms must be applied to the information sensed by the radar. From actual data collected by radar the stages and algorithms needed to obtain ISAR images are revised, including high resolution range profile generation, motion compensation and ISAR formation. Target recognition is achieved by comparing the generated set of actual ISAR images with a database of ISAR images generated by electromagnetic software. High resolution radar image generation and target recognition processes are burdensome and time consuming, so to determine the most suitable implementation platform the analysis of the computational complexity is of great interest. To this end and since target identification must be completed in real time, computational burden of both processes the generation and comparison with a database is explained separately. Conclusions are drawn about implementation platforms and calculation efficiency in order to reduce time consumption in a possible future implementation.

Low-Cost CW-LFM Radar Sensor at 100 GHz

This paper presents a <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">W</i> -band high-resolution radar sensor for short-range applications. Low-cost technologies have been properly selected in order to implement a versatile and easily scalable radar system. A large operational bandwidth of 9 GHz, required for obtaining high-range resolution, is attained by means of a frequency multiplication-based architecture. The system characterization to identify the performance-limiting stages and the subsequent design optimization are presented. The assessment of system performance for several representative applications has been carried out.

DSM GENERATION FROM HIGH RESOLUTION COSMO-SKYMED IMAGERY WITH RADARGRAMMETRIC MODEL

Abstract. The availability of new high resolution radar spaceborne sensors offers new interesting potentialities for the geomatics application: spatial and temporal change detection, features extraction, generation of Digital Surface (DSMs). As regards the DSMs generation from new high resolution data (as SpotLight imagery), the development and the accuracy assessment of method based on radargrammetric approach are topics of great interest and relevance. The aim of this investigation is the DSM generation from a COSMO-SkyMed Spotlight stereo pair with the radargrammetric technique. DSM generation procedure consists of two basic steps: the stereo pair orientation and the image matching. The suite for radargrammetric approach has been implemented in SISAR (Software per Immagini Satellitari ad Alta Risoluzione), a scientific software developed at the Geodesy and Geomatic Institute of the University of Rome "La Sapienza". As regard the image matching the critical issue is the definition of a strategy to search the corresponding points; in SISAR software, an original matching procedure has been developed, based on a coarse-to-fine hierarchical solution with an effective combination of geometrical constrains and an Area Base Matching (ABM) algorithm.

High-Resolution SAR Radargrammetry: A First Application With COSMO-SkyMed SpotLight Imagery

The availability of new high-resolution radar spaceborne sensors offers new interesting potentialities for the acquisition of data useful for the generation of Digital Surface Models (DSMs). Two different approaches may be used to generate DSMs from Synthetic Aperture Radar (SAR) data: the interferometric and the radargrammetric one. At present, the importance of the radargrammetric approach is rapidly growing due to the new high-resolution imagery [up to 1 m Ground Sample Distance (GSD)] which can be acquired by COSMO-SkyMed, TerraSAR-X and RADARSAT-2 in SpotLight mode. The defined and implemented model is related to COSMO- SkyMed SpotLight imagery in zero-Doppler geometry; it performs a 3-D orientation based on two range and two zero-Doppler equations, allowing for the least squares estimation of some calibration parameters, related to satellite position and velocity and to the range measure. The model has been implemented in SISAR (Software per Immagini Satellitari ad Alta Risoluzione), a scientific software developed at the Geodesy and Geomatic Institute of the University of Rome “La Sapienza”. Starting from this model, based on a geometric reconstruction, also a tool for the Rational Polynomial Coefficients (RPCs) generations has been implemented. To test the effectiveness of the new model, a stereo pair over the test sites of Merano (Northern Italy) has been orientated using the rigorous model and the RPCs one, and first results of radargrammetric DSM generation are presented; they display the possibility to reach an overall average accuracy of 3.5 m.

Millimeter-Wave Sensor With FMICW Capabilities for Medium-Range High-Resolution Radars

Frequency-modulated continuous-wave (FMCW) radars offer many advantages such as low peak power, low probability of interception, low interference with other systems, and high-range resolution. However, their major drawback is the isolation required between the transmitter and receiver that typically leads to the use of two separate antennas for transmission and reception. Some schemes have been proposed to enable FMCW radars to operate with a single antenna, such as frequency-modulated interrupted continuous-wave (FMICW) technology. Recently, a stagger procedure has been proposed to overcome the problems associated with the use of the FMICW technique in high-resolution radars. The technique was tested using simulated data. Now, a high-resolution millimeter-wave radar sensor, in <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Ka</i> -band, has been developed at the Universidad Politecnica de Madrid, Madrid, Spain, to perform an experimental validation of the theoretical approach of the stagger procedure. The sensor transmits a linear frequency-modulated interrupted continuous wave with a maximum bandwidth of 2 GHz and a transmitted power of +29 dBm.

Integrated object and road border tracking using 77 GHz automotive radars

Radar sensors are widely used in automotive on-board systems such as adaptive cruise control, which tracks the preceding vehicle and keeps a safe distance. The authors go a step further and exploit the potential of a high resolution radar sensor for the estimation of the road borders and the reconstruction of the traffic scenario based on object tracking. A FMCW millimetre wave radar is mounted in the front bumper of a vehicle and delivers object lists to the central fusion processor, where detections are considered as measurements to an integrated model based road border estimation system. The algorithm presented is designed to improve the prediction of paths and trajectories of the ego-vehicle in future enhanced collision warning and collision avoidance systems. Given the characteristics and limitations of a radar system, such as heavy clutter, the algorithms are tested with real data and perform reliably in highway and extra-urban scenarios and are competitive with data fusion and stereo vision systems, but with lower computational load.