Pedestrian Classification for 79 GHz Automotive Radar Systems

Prophet, Robert; Hoffmann, Marcel; Ossowska, Alicja; Malik, Waqas Ahmed; Sturm, Christian; Vossiek, Martin

doi:10.1109/ivs.2018.8500554

Cited by 30 publications

(14 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…First applied on SAR data, classification algorithms have been recently applied to distinguish between various types of targets detected by radar [117]- [120]. A special case of target identification that has attracted particular interest is the recognition of vulnerable road users (VRU) such as pedestrians, either based on µDoppler signatures [121]- [123], range-velocity spectra [124], [125], or other [126]. Besides classification of specific targets, deep learning can provide better scene understanding by semantic segmentation [127].…”

Section: E Machine Learning and Automotive Radarmentioning

confidence: 99%

Automotive Radar — From First Efforts to Future Systems

2021

View full text Add to dashboard Cite

Section: E Machine Learning and Automotive Radarmentioning

confidence: 99%

Automotive Radar — From First Efforts to Future Systems

2021

View full text Add to dashboard Cite

“…However, despite this approach is well-suited for static-objects, it fails on dynamic ones, as the object velocity needs to be tracked to avoid generating a tail of reflections. Another classical radar processing approach consists in three steps: clustering, extraction of hand-crafted features and classification [20]. Wöhler et al [25] designed a method which uses DBSCAN [7] at the clustering stage.…”

Section: Deep Learning On Radar Datamentioning

confidence: 99%

A Neural Network Based System for Efficient Semantic Segmentation of Radar Point Clouds

Cennamo

Kaestner

Kummert

2021

Neural Process Lett

View full text Add to dashboard Cite

The last decade has witnessed important advancements in the field of computer vision and scene understanding, enabling applications such us autonomous vehicles. Radar is a commonly adopted sensor in automotive industry, but its suitability to machine learning techniques still remains an open question. In this work, we propose a neural network (NN) based solution to efficiently process radar data. We introduce RadarPCNN, an architecture specifically designed for performing semantic segmentation on radar point clouds. It uses PointNet$$++$$ + + as a building-block—enhancing the sampling stage with mean-shift—and an attention mechanism to fuse information. Additionally, we propose a machine learning radar pre-processing module that confers the network the ability to learn from radar features. We show that our solutions are effective, yielding superior performance than the state-of-the-art.

show abstract

“…In the other previous research, the moving pedestrian was detected, tracked, and its shape was measured by using four-dimensional radar processing and particle filter applying to detected point cloud [8]. Furthermore, machine learning was used for pedestrian classification in 79 GHz automotive radar systems [9]. However, the amount of calculation in these methods increases according to the number of point clouds and not suitable for real-time and cost performance required to personal mm-wave radar device.…”

Section: Introductionmentioning

confidence: 99%

An experimental study of obstacles and pedestrian classification by relative velocity distribution on urban-narrow bicycle-pedestrian road with 77 GHz FMCW radar

et al. 2021

View full text Add to dashboard Cite

In recent years, the number of people using the bicycle as a means of transportation has increased due to heightened health awareness and environmental protection and COVID-19 pandemic, instead of public such as train and bus. Further, bicycle-related traffic accidents have rapidly increased due to the rapid growth of the food-delivery business with bicycle use, and thus countermeasures are urgently needed. In this paper, we propose a method for a bicycle to detect pedestrian separating from road obstacles such as guard poles and guardrails installed in an urban-narrow bicycle-pedestrian road with the use of frequency modulated continuous wave multi-input multi-output (FMCW-MIMO) radar. In the proposed method, from the distribution of the relative velocity of the detected objects, the objects gathering around the mode of the relative-velocity distribution are judged as the road obstacles, and the objects far from the mode are judged as pedestrian. In an experiment, a bicycle equipped with 77 GHzband FMCW-MIMO radar, ran at a speed of 2 m/s and a pedestrian walked ahead at a speed of 1 m/s in a narrow bicycle-pedestrian road (school zone) in central Tokyo. Experimental result shows that the pedestrian was detected with about 1m/s-higher relative velocity than around 1.5 m/s of mode relative velocity, where road obstacles group distributed in the range of ± 0.5 m/s.

show abstract

Pedestrian Classification for 79 GHz Automotive Radar Systems

Cited by 30 publications

References 8 publications

Automotive Radar — From First Efforts to Future Systems

Automotive Radar — From First Efforts to Future Systems

A Neural Network Based System for Efficient Semantic Segmentation of Radar Point Clouds

An experimental study of obstacles and pedestrian classification by relative velocity distribution on urban-narrow bicycle-pedestrian road with 77 GHz FMCW radar

Contact Info

Product

Resources

About