Abstract:Advanced driver assistance systems (ADASs) and autonomous vehicles rely on different types of sensors, such as camera, radar, ultrasonic, and LiDAR, to sense the surrounding environment. Compared with the other types of sensors, millimeter-wave automotive radar has advantages in terms of cost and reliability under bad weather conditions (e.g., snow, rain, and fog) and does not suffer from light condition variations (e.g., darkness). Typical radar devices used in today's commercial vehicles with ADAS features produce sparse point clouds in low angular resolution with a limited number of antennas. In this article, we present a machine-learning-aided signal processing chain to suppress the radar imaging blur effect introduced by the phase migration in time-division multiplexing multiple-input multiple-output radar, to generate low-level high-resolution radar bird's-eye view (BEV) spectra with rich object's features. Compared with radar point clouds, there is no information loss in radar BEV spectra. We then propose a temporal-fusion distance-tolerant single-stage object detection network, termed as TDRadarNet, and an enhanced version, TDRadarNet+, to robustly detect vehicles in both long and short ranges on radar BEVs. We introduce a first-of-its-kind multimodel dataset, containing 14 800 frames of high-resolution low-level radar BEV spectra with synchronized stereo camera RGB images and 3-D LiDAR point clouds. Our dataset achieves 0.39-m range resolution and 1.2 degrees degree azimuth angular resolution with 100-m maximum detectable range. Moreover, we create a subdataset, the Doppler Unfolding dataset, containing 244 140 beam vectors extracted from the 3-D radar data cube. With extensive testing and evaluation, we demonstrate that our Doppler unfolding network achieves 93.46% Doppler unfolding accuracy. Compared to YOLOv7, a state-of-the-art image-based object detection network, TDRadarNet, achieves a 70.3% average precision (AP) for vehicle detection, demonstrating a 21.0% improvement; TDRadarNet+ achieves a 73.9% AP, showing a 24.6% improvement in performance.

Automatic Multipath Annotation for Conventional Automotive Radar Datasets

A Deep Automotive Radar Detector using the RaDelft Dataset

A New Automotive Radar 4D Point Clouds Detector by Using Deep Learning

Dual Radar: A Multi-modal Dataset with Dual 4D Radar for Autononous Driving

DeepHybrid: Deep Learning on Automotive Radar Spectra and Reflections for Object Classification

RADDet: Range-Azimuth-Doppler based Radar Object Detection for Dynamic Road Users

Estimating the Magnitude and Phase of Automotive Radar Signals Under Multiple Interference Sources With Fully Convolutional Networks

Raw High-Definition Radar for Multi-Task Learning

Towards Deep Radar Perception for Autonomous Driving: Datasets, Methods, and Challenges

Dual Radar: A Multi-modal Dataset with Dual 4D Radar for Autonomous Driving

Deep Stochastic Radar Models

The Radar Ghost Dataset -- An Evaluation of Ghost Objects in Automotive Radar Data

Beyond Point Clouds: A Knowledge-Aided High Resolution Imaging Radar Deep Detector for Autonomous Driving

Concept for an Automatic Annotation of Automotive Radar Data Using AI-segmented Aerial Camera Images

Semi-Automatic Annotation of 3D Radar and Camera for Smart Infrastructure-Based Perception

See Further Than CFAR: a Data-Driven Radar Detector Trained by Lidar

Deep-Neural-Network-Enabled Vehicle Detection Using High-Resolution Automotive Radar Imaging

Automotive Radar Signal Processing: Research Directions and Practical Challenges

Efficient 4D Radar Data Auto-labeling Method using LiDAR-based Object Detection Network

Deep Radar Detector

Spectranet: A High Resolution Imaging Radar Deep Neural Network for Autonomous Vehicles.