Radar-based 2D Car Detection Using Deep Neural Networks

Abstract

A crucial part of safe navigation of autonomous vehicles is the robust detection of surrounding objects. While there are numerous approaches covering object detection in images or LiDAR point clouds, this paper addresses the problem of object detection in radar data. For this purpose, the fully convolutional neural network YOLOv3 is adapted to operate on sparse radar point clouds. In order to apply convolutions, the point cloud is transformed into a grid-like structure. The impact of this representation transformation is shown by comparison with a network based on Frustum PointNets, which directly processes point cloud data. The presented networks are trained and evaluated on the public nuScenes dataset. While experiments show that the point cloud-based network outperforms the grid-based approach in detection accuracy, the latter has a significantly faster inference time neglecting the grid conversion which is crucial for applications like autonomous driving.

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

“…It is necessary to increase the input density of the radar point cloud for better performance. Dreher et al [141] accumulated radar points into an occupancy grid mapping (OGM), then applied YOLOv3 [142] for object detection. Some works [143][144][145] utilise point cloud segmentation networks, such as PointNet [146] and PointNet++ [147], followed by a bounding box regression module for 2D radar detection.…”

Self-Supervised Velocity Estimation for Automotive Radar Object Detection Networks

“…These point cloud based networks can be further differentiated into grid-based and point-based architectures. Grid-based approaches first render the point cloud into a 2D bird eye view (BEV) or 3D voxel grid using hand-crafted operations [11], [28], [29], [30], [31] or learned feature-encoders [32], [12], [31] and subsequently apply convolutional backbones to the grid.…”

“…detection networks are still relatively new for radar, whereas they can already be considered state-of-the-art for image processing [2], [4] and lidar [3], [5]. In recent research, radar data, mainly in the form of spectra or preprocessed point clouds, are used as input to a neural network to solve various problems such as semantic segmentation [6], [7], classification [8], [9], object detection [1], [10], [11], [12], [13] or tracking [14], [15].…”

Evaluation of the bridge reinforcement effect based on vehicle-bridge coupling vibration

“…Similarly, clustering with semantic information is utilized in [29] for radar point cloud instance segmentation, and the same approach is combined with contrastive learning in [30] for solving lacking of radar point annotations. Moreover, a PointNet-based network is employed in [4] for object classification, segmentation, and 2D bounding box prediction on the bird's-eye view (BEV), whereas 3D bounding box prediction is addressed in [18].…”

“…sidered to be directly extended for 4D radar-based object detection directly. For instance, researchers have proposed PointNet-based [4][17] [18] or graph-convolution-based [5] [19][20] methods as potential solutions. These end-to-end learning methods offer advantages such as larger receptive fields and simplified structures.…”