Pedestrian classification with 24 GHz chirp sequence radar

Sorowka, Peter; Rohling, Hermann

doi:10.1109/irs.2015.7226392

Cited by 12 publications

(4 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Due to the high complexity of identifying object instances in these kinds of data, most of the research utilizes radar for gaining a basic first understanding about the scene at hand. Hence, only few classes are separated from each other, e.g., [2]- [4]. Excellent environmental perception requires understanding more complex scenes.…”

Section: Introductionmentioning

confidence: 99%

Radar-based Road User Classification and Novelty Detection with Recurrent Neural Network Ensembles

Scheiner

Appenrodt

Dickmann

et al. 2019

2019 IEEE Intelligent Vehicles Symposium (IV)

View full text Add to dashboard Cite

Radar-based road user classification is an important yet still challenging task towards autonomous driving applications. The resolution of conventional automotive radar sensors results in a sparse data representation which is tough to recover by subsequent signal processing. In this article, classifier ensembles originating from a one-vs-one binarization paradigm are enriched by one-vs-all correction classifiers. They are utilized to efficiently classify individual traffic participants and also identify hidden object classes which have not been presented to the classifiers during training. For each classifier of the ensemble an individual feature set is determined from a total set of 98 features. Thereby, the overall classification performance can be improved when compared to previous methods and, additionally, novel classes can be identified much more accurately. Furthermore, the proposed structure allows to give new insights in the importance of features for the recognition of individual classes which is crucial for the development of new algorithms and sensor requirements.

show abstract

Section: Introductionmentioning

confidence: 99%

Radar-based Road User Classification and Novelty Detection with Recurrent Neural Network Ensembles

Scheiner

Appenrodt

Dickmann

et al. 2019

2019 IEEE Intelligent Vehicles Symposium (IV)

View full text Add to dashboard Cite

show abstract

“…However, the use of automotive three-dimensional (3D) radar for target detection and tracking on water is novel. Automotive radar has been proved to work for some onshore tasks, including pedestrian tracking [30][31][32][33] and terrain mapping [34,35]. Radar technology is often also used as part of a larger combined system for navigational purposes, for example in 3D mapping [36], sensor fusion [37], and navigational decision support systems [38][39][40].…”

Section: Problem Definition and Major Contributionmentioning

confidence: 99%

Shore Construction Detection by Automotive Radar for the Needs of Autonomous Surface Vehicle Navigation

Stateczny

Kazimierski

Burdziakowski

et al. 2019

IJGI

View full text Add to dashboard Cite

Autonomous surface vehicles (ASVs) are becoming more and more popular for performing hydrographic and navigational tasks. One of the key aspects of autonomous navigation is the need to avoid collisions with other objects, including shore structures. During a mission, an ASV should be able to automatically detect obstacles and perform suitable maneuvers. This situation also arises in near-coastal areas, where shore structures like berths or moored vessels can be encountered. On the other hand, detection of coastal structures may also be helpful for berthing operations. An ASV can be launched and moored automatically only if it can detect obstacles in its vicinity. One commonly used method for target detection by ASVs involves the use of laser rangefinders. The main disadvantage of this approach is that such systems perform poorly in conditions with bad visibility, such as in fog or heavy rain. Therefore, alternative methods need to be sought. An innovative approach to this task is presented in this paper, which describes the use of automotive three-dimensional radar on a floating platform. The goal of the study was to assess target detection possibilities based on a comparison with photogrammetric images obtained by an unmanned aerial vehicle (UAV). The scenarios considered focused on analyzing the possibility of detecting shore structures like berths, wooden jetties, and small houses, as well as natural objects like trees or other kinds of vegetation. The recording from the radar was integrated into a single complex radar image of shore targets. It was then compared with an orthophotomap prepared from AUV camera pictures, as well as with a map based on traditional land surveys. The possibility and accuracy of detection for various types of shore structure were statistically assessed. The results show good potential for the proposed approach—in general, objects can be detected using the radar—although there is a need for development of further signal processing algorithms.

show abstract

“…Other studies looked at using range‐Doppler maps as the domain to perform classification. Object tracking through clustering algorithms and a linear classifier was used to distinguish vehicles and scenarios of walking pedestrians in [9], and in [10] features related to the size, orientation, and frequency of the pedestrians’ step were used in conjunction with ordered statistics‐constant false alarm rate (OS‐CFAR) and density‐based cluster algorithm. Further works focused on using different domains of information to achieve vehicles–pedestrians classification, such as [11] through the phase characteristics (coherent/non‐coherent) of the object signature, and [12] through features related to the differences in radar cross section (RCS) between the different classes of targets, used together with a support vector machine classifier.…”

Section: Introductionmentioning

confidence: 99%

Practical classification of different moving targets using automotive radar and deep neural networks

Angelov

Robertson

Murray-Smith

et al. 2018

IET Radar, Sonar & Navigation

104

View full text Add to dashboard Cite

In this work, we present results for classification of different classes of targets (car, single and multiple people, bicycle) using automotive radar data and different neural networks. A fast implementation of radar algorithms for detection, tracking, and micro-Doppler extraction is proposed in conjunction with the automotive radar transceiver TEF810X and microcontroller unit SR32R274 manufactured by NXP Semiconductors. Three different types of neural networks are considered, namely a classic convolutional network, a residual network, and a combination of convolutional and recurrent network, for different classification problems across the 4 classes of targets recorded. Considerable accuracy (close to 100% in some cases) and low latency of the radar pre-processing prior to classification (approximately 0.55s to produce a 0.5s long spectrogram) are demonstrated in this paper, and possible shortcomings and outstanding issues are discussed.

show abstract

Pedestrian classification with 24 GHz chirp sequence radar

Cited by 12 publications

References 3 publications

Radar-based Road User Classification and Novelty Detection with Recurrent Neural Network Ensembles

Radar-based Road User Classification and Novelty Detection with Recurrent Neural Network Ensembles

Shore Construction Detection by Automotive Radar for the Needs of Autonomous Surface Vehicle Navigation

Practical classification of different moving targets using automotive radar and deep neural networks

Contact Info

Product

Resources

About