The volcanormal density for radar-based extended target tracking

Broseit, Peter; Duraisamy, Bharanidhar; Dickmann, Jürgen

doi:10.1109/itsc.2017.8317736

Cited by 21 publications

(13 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The ratio of core points to the total number of detections serves as additional feature. Inspired by the findings in [14], the resemblance of the detection distribution to an ellipse shape is estimated. Therefore, the ratio of points close to the estimated outlines of the object to the total amount of detections is calculated.…”

Section: A Ensemble Creation With Classifier Specific Feature Setsmentioning

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)

Self Cite

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: A Ensemble Creation With Classifier Specific Feature Setsmentioning

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)

Self Cite

View full text Add to dashboard Cite

show abstract

“…A prominent example is the elliptical random matrix model [18] which has been extended for incorporating Doppler measurements in [19]. Also, a polynomial object model for tracking stationary objects such as guard rails [20] and a Volcanormal density for modeling vehicles [21] were proposed for radar applications.…”

Section: Introductionmentioning

confidence: 99%

Tracking Multiple Vehicles Using a Variational Radar Model

Scheel

Dietmayer

2019

IEEE Trans. Intell. Transport. Syst.

View full text Add to dashboard Cite

High-resolution radar sensors are able to resolve multiple detections per object and therefore provide valuable information for vehicle environment perception. For instance, multiple detections allow to infer the size of an object or to more precisely measure the object's motion. Yet, the increased amount of data raises the demands on tracking modules: measurement models that are able to process multiple detections for an object are necessary and measurement-to-object associations become more complex. This paper presents a new variational radar model for tracking vehicles using radar detections and demonstrates how this model can be incorporated into a Random-Finite-Setbased multi-object filter. The measurement model is learned from actual data using variational Gaussian mixtures and avoids excessive manual engineering. In combination with the multiobject tracker, the entire process chain from the raw measurements to the resulting tracks is formulated probabilistically. The presented approach is evaluated on experimental data and it is demonstrated that the data-driven measurement model outperforms a manually designed model.

show abstract

“…with the orientations α x1 and αx 2 restricted as in (8). The sums can be simplified using the 2π periodicity of the orientation to reduce the number of components to the 4 most likely ones.…”

Section: A Explicit Parameter Spacementioning

confidence: 99%

“…back. Due to the aforementioned ambiguity in ellipse parameterization, we utilize the wrapped distribution from (8). To make the calculation feasible, we use the 2π periodicity of the orientation to reduce the components to the 4 most likely ones similar to the MWDP, meaning the Gaussian likelihood is calculated forx with j ∈ {0, 1, 2, 3} and the sum is used to weight the particlẽ…”

Section: B Transformed Spacementioning

confidence: 99%

Fusion of Elliptical Extended Object Estimates Parameterized With Orientation and Axes Lengths

Thormann

Baum

2021

IEEE Trans. Aerosp. Electron. Syst.

View full text Add to dashboard Cite

This article considers the fusion of target estimates stemming from multiple sensors, where the spatial extent of the targets is incorporated. The target estimates are represented as ellipses parameterized with center orientation and semi-axis lengths and width. Here, the fusion faces challenges such as ambiguous parameterization and an unclear meaning of the Euclidean distance between such estimates. We introduce a novel Bayesian framework for random ellipses based on the concept of a Minimum Mean Gaussian Wasserstein (MMGW) estimator. The MMGW estimate is optimal with respect to the Gaussian Wasserstein (GW) distance, which is a suitable distance metric for ellipses. We develop practical algorithms to approximate the MMGW estimate of the fusion result. The key idea is to approximate the GW distance with a modified version of the Square Root (SR) distance. By this means, optimal estimation and fusion can be performed based on the square root of the elliptic shape matrices. We analyze different implementations using, e.g., Monte Carlo methods, and evaluate them in simulated scenarios. An extensive comparison with state-of-the-art methods highlights the benefits of estimators tailored to the Gaussian Wasserstein distances.

show abstract

The volcanormal density for radar-based extended target tracking

Cited by 21 publications

References 8 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

Tracking Multiple Vehicles Using a Variational Radar Model

Fusion of Elliptical Extended Object Estimates Parameterized With Orientation and Axes Lengths

Contact Info

Product

Resources

About