Deep Instance Segmentation With Automotive Radar Detection Points

Liu, Jianan; Xiong, Weiyi; Bai, Liping; Xia, Yuxuan; Huang, Tao; Ouyang, Wanli; Zhu, Bing

doi:10.1109/tiv.2022.3168899

Cited by 42 publications

(9 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Experiments showed that the OGM-based method performs best, while the PointNet-based method performs far worse than others probably due to sparsity. Liu et al [148] suggest that incorporating global information can help with the sparsity issue of the radar point cloud. Therefore, they added a gMLP [149] block to each set abstraction layer in PointNet++.…”

Section: Point Cloud Detectormentioning

confidence: 99%

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

Zhou

Liu

Zhao

et al. 2022

Sensors

View full text Add to dashboard Cite

With recent developments, the performance of automotive radar has improved significantly. The next generation of 4D radar can achieve imaging capability in the form of high-resolution point clouds. In this context, we believe that the era of deep learning for radar perception has arrived. However, studies on radar deep learning are spread across different tasks, and a holistic overview is lacking. This review paper attempts to provide a big picture of the deep radar perception stack, including signal processing, datasets, labelling, data augmentation, and downstream tasks such as depth and velocity estimation, object detection, and sensor fusion. For these tasks, we focus on explaining how the network structure is adapted to radar domain knowledge. In particular, we summarise three overlooked challenges in deep radar perception, including multi-path effects, uncertainty problems, and adverse weather effects, and present some attempts to solve them.

show abstract

Section: Point Cloud Detectormentioning

confidence: 99%

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

Zhou

Liu

Zhao

et al. 2022

Sensors

View full text Add to dashboard Cite

show abstract

“…The stored information is used to calculate additional point features and improve segmentation accuracy. Other modifications of PointNet++, such as replacing the sampling step with mean shift clustering or adding more processing blocks, are described in [21] and [22]. Approaches for the semantic segmentation of moving objects that are not based on PointNet++ are [23] and [24].…”

Section: Related Workmentioning

confidence: 99%

Fast Rule-Based Clutter Detection in Automotive Radar Data

Kopp

Kellner

Piroli

et al. 2021

2021 IEEE International Intelligent Transportation Systems Conference (ITSC)

View full text Add to dashboard Cite

The unique properties of radar sensors, such as their robustness to adverse weather conditions, make them an important part of the environment perception system of autonomous vehicles. One of the first steps during the processing of radar point clouds is often the detection of clutter, i.e. erroneous points that do not correspond to real objects. Another common objective is the semantic segmentation of moving road users. These two problems are handled strictly separate from each other in literature. The employed neural networks are always focused entirely on only one of the tasks. In contrast to this, we examine ways to solve both tasks at the same time with a single jointly used model. In addition to a new augmented multi-head architecture, we also devise a method to represent a network's predictions for the two tasks with only one output value. This novel approach allows us to solve the tasks simultaneously with the same inference time as a conventional task-specific model. In an extensive evaluation, we show that our setup is highly effective and outperforms every existing network for semantic segmentation on the RadarScenes dataset [1].

show abstract

“…At present, high-resolution lidar measurement technology plays an important role in fields such as ocean exploration [ 1 ] and atmospheric environment monitoring [ 2 ]. Laser micro-Doppler technology, as one of the high-resolution measurement methods, has been widely used in various fields such as vital sign detection [ 3 , 4 , 5 , 6 ], drone detection [ 7 , 8 ], automotive radar detection [ 9 ], vibration measurement [ 10 ], and bird recognition [ 11 ]. When a radar emits electromagnetic waves onto the surface of a moving object, the echo frequency will undergo a Doppler shift compared to the emitting signal.…”

Section: Introductionmentioning

confidence: 99%

Decoupling and Parameter Extraction Methods for Conical Micro-Motion Object Based on FMCW Lidar

Yang,

Liu

et al. 2024

Sensors

View full text Add to dashboard Cite

Micro-Doppler time–frequency analysis has been regarded as an important parameter extraction method for conical micro-motion objects. However, the micro-Doppler effect caused by micro-motion can modulate the frequency of lidar echo, leading to coupling between structure and micro-motion parameters. Therefore, it is difficult to extract parameters for micro-motion cones. We propose a new method for parameter extraction by combining the range profile of a micro-motion cone and the micro-Doppler time–frequency spectrum. This method can effectively decouple and accurately extract the structure and the micro-motion parameters of cones. Compared with traditional time–frequency analysis methods, the accuracy of parameter extraction is higher, and the information is richer. Firstly, the range profile of the micro-motion cone was obtained by using an FMCW (Frequency Modulated Continuous Wave) lidar based on simulation. Secondly, quantitative analysis was conducted on the edge features of the range profile and the micro-Doppler time–frequency spectrum. Finally, the parameters of the micro-motion cone were extracted based on the proposed decoupling parameter extraction method. The results show that our method can effectively extract the cone height, the base radius, the precession angle, the spin frequency, and the gravity center height within the range of a lidar LOS (line of sight) angle from 20° to 65°. The average absolute percentage error can reach below 10%. The method proposed in this paper not only enriches the detection information regarding micro-motion cones, but also improves the accuracy of parameter extraction and establishes a foundation for classification and recognition. It provides a new technical approach for laser micro-Doppler detection in accurate recognition.

show abstract

Deep Instance Segmentation With Automotive Radar Detection Points

Cited by 42 publications

References 25 publications

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

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

Fast Rule-Based Clutter Detection in Automotive Radar Data

Decoupling and Parameter Extraction Methods for Conical Micro-Motion Object Based on FMCW Lidar

Contact Info

Product

Resources

About