Search

Your search keyword '"Kellner, Dominik"' showing total 31 results
Start Over Author "Kellner, Dominik"
31 results on '"Kellner, Dominik"'

1. Simultaneous Clutter Detection and Semantic Segmentation of Moving Objects for Automotive Radar Data

Author

Kopp, Johannes, Kellner, Dominik, Piroli, Aldi, Dallabetta, Vinzenz, and Dietmayer, Klaus

Subjects
Computer Science - Computer Vision and Pattern Recognition, Computer Science - Machine Learning, Electrical Engineering and Systems Science - Signal Processing
Abstract

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., Comment: Published at IEEE International Conference on Intelligent Transportation Systems (ITSC), Bilbao, ESP, 2023

Published
2023

2. Tackling Clutter in Radar Data -- Label Generation and Detection Using PointNet++

Author

Kopp, Johannes, Kellner, Dominik, Piroli, Aldi, and Dietmayer, Klaus

Subjects
Computer Science - Computer Vision and Pattern Recognition, Computer Science - Machine Learning, Computer Science - Robotics, Electrical Engineering and Systems Science - Signal Processing
Abstract

Radar sensors employed for environment perception, e.g. in autonomous vehicles, output a lot of unwanted clutter. These points, for which no corresponding real objects exist, are a major source of errors in following processing steps like object detection or tracking. We therefore present two novel neural network setups for identifying clutter. The input data, network architectures and training configuration are adjusted specifically for this task. Special attention is paid to the downsampling of point clouds composed of multiple sensor scans. In an extensive evaluation, the new setups display substantially better performance than existing approaches. Because there is no suitable public data set in which clutter is annotated, we design a method to automatically generate the respective labels. By applying it to existing data with object annotations and releasing its code, we effectively create the first freely available radar clutter data set representing real-world driving scenarios. Code and instructions are accessible at www.github.com/kopp-j/clutter-ds., Comment: To be published in IEEE International Conference on Robotics and Automation (ICRA), London, UK, 2023

Published
2023

3. Fast Rule-Based Clutter Detection in Automotive Radar Data

Author

Kopp, Johannes, Kellner, Dominik, Piroli, Aldi, and Dietmayer, Klaus

Subjects
Computer Science - Computer Vision and Pattern Recognition, Electrical Engineering and Systems Science - Signal Processing
Abstract

Automotive radar sensors output a lot of unwanted clutter or ghost detections, whose position and velocity do not correspond to any real object in the sensor's field of view. This poses a substantial challenge for environment perception methods like object detection or tracking. Especially problematic are clutter detections that occur in groups or at similar locations in multiple consecutive measurements. In this paper, a new algorithm for identifying such erroneous detections is presented. It is mainly based on the modeling of specific commonly occurring wave propagation paths that lead to clutter. In particular, the three effects explicitly covered are reflections at the underbody of a car or truck, signals traveling back and forth between the vehicle on which the sensor is mounted and another object, and multipath propagation via specular reflection. The latter often occurs near guardrails, concrete walls or similar reflective surfaces. Each of these effects is described both theoretically and regarding a method for identifying the corresponding clutter detections. Identification is done by analyzing detections generated from a single sensor measurement only. The final algorithm is evaluated on recordings of real extra-urban traffic. For labeling, a semi-automatic process is employed. The results are promising, both in terms of performance and regarding the very low execution time. Typically, a large part of clutter is found, while only a small ratio of detections corresponding to real objects are falsely classified by the algorithm., Comment: To be published in IEEE 24th International Conference on Intelligent Transportation Systems (ITSC), Indianapolis, USA, 2021

Published
2021

9. Verfahren zur Bestimmung von Objekt- und Eigenbewegung auf Basis der Dopplerinformation hochauflösender Radarsensoren

Author

Kellner, Dominik, Dietmayer, Klaus, and Waldschmidt, Christian

Subjects
Radar, Driver assistance systems, Tracking, Fahrerassistenzsystem, Doppler radar, DDC 620 / Engineering & allied operations, Doppler-Radar, ddc:620, Objektverfolgung
Abstract

Im Bereich der Umfelderfassung von Fahrerassistenzsystemen wird eine immer größere Anzahl an Radarsensoren eingesetzt. Zudem nimmt die Messgenauigkeit kontinuierlich zu, im Winkel durch digitale Strahlformung und hochauflösende Algorithmen und im Doppler durch z.B. das Schnelle-Rampen-Verfahren. In Verbindung ergeben sich dadurch neue Möglichkeiten bei der Bestimmung der Eigenbewegung und beim Erfassen ausgedehnter Ziele im Fahrzeugumfeld. Im Rahmen der Arbeit wird der Verlauf der Dopplergeschwindigkeit über dem Azimutwinkel eines hochauflösenden Automotive-Radars analysiert. Dieser charakteristische Verlauf eines ausgedehnten Ziels (bzw. mehrerer stationärer Ziele) wird mit Geschwindigkeitsprofil bezeichnet und kann in direkten Zusammenhang zu dem aktuellen Bewegungszustand gesetzt werden. Es werden dazu Methoden vorgestellt, die das Geschwindigkeitsprofil erstmalig mittels eines optimalen, echtzeitfähigen und erwartungstreuen Schätzers bestimmen. Wird ein starres, ausgedehntes Ziel (z.B. Fahrzeug) von zwei Radarsensoren erfasst, kann durch Kombination der Geschwindigkeitsprofile instantan, das heißt in einem Messzyklus, sein vollständiger Bewegungszustand (Geschwindigkeit, Gierrate, Drift) bestimmt werden. Im Gegensatz zu aktuell angewendeten Tracking-Verfahren kann dabei vollständig auf Modellannahmen (Bewegungsmodell) und eine zeitliche Filterung verzichtet werden. Ferner benötigt das Verfahren keine Initialisierungsphase und hat keine Latenz bei Bewegungsänderungen. Der Zeitgewinn und die gesteigerte Genauigkeit sind für zukünftige Fahrerassistenzsysteme ein entscheidender Vorteil, um deutlich schneller auf kritische Verkehrssituationen reagieren zu können. Wertet man das Geschwindigkeitsprofil aller erfassten stationären Ziele aus, kann in einem einzelnen Messzyklus der Bewegungszustand des eigenen Fahrzeugs bestimmt werden. Dazu wird ein Verfahren entwickelt, bei dem beliebig viele und beliebig angebrachte Sensoren kombiniert werden. Die stationären Ziele werden direkt durch ein robustes Schätzverfahren identifiziert. Insbesondere weist die Schätzung der Geschwindigkeit eine sehr hohe Genauigkeit auf. Bei mehreren Sensoren kann sogar ein lateraler Drift der Räder bestimmt werden. Zudem ist das Verfahren im Gegensatz zu der Fahrzeugodometrie (Gyroskope, Raddrehzahlsensoren) driftfrei, weist keinen Skalierungsfehler auf und liefert auch bei instabilen Fahrzuständen einen präzisen Bewegungszustand. Wird die radarbasierte Eigenbewegungsschätzung mit der Fahrzeugodometrie kombiniert, ist es möglich die Einbauorientierung des Radarsensors online und sehr präzise zu bestimmen. Zudem kann der Bias-Fehler und Skalierungsfehler der Fahrzeugodometrie in Echtzeit korrigiert werden. Die hohe Abtastrate und Genauigkeit des Gyroskops wird dadurch mit der radarbasierten Eigenbewegungsschätzung, die frei von systematischen Fehlern ist, kombiniert., A growing number of Doppler radar sensors are being implemented in the environmental perception of advanced driving assistance systems. Furthermore, their measurement accuracy in Doppler estimation and angle determination using digital beamforming, chirp sequence modulation and high-resolution algorithms is continuously increasing. In combination, this results in new possibilities for estimating the motion state of the ego-vehicle and of extended objects in the vehicle's environment. Within the scope of this thesis, the progression of the Doppler velocity over the azimuth angle of a high-resolution radar is analyzed. Referred to as the velocity profile, this characteristic progression of an extended object (or multiple stationary objects) can be related directly to the current motion state. A method is derived which for the first time allows determination of the velocity profile using an optimal, real-time capable and unbiased estimator. If a rigid, extended object (e.g. a vehicle) is detected by two radar sensors, an estimate of the full motion state (velocity, yaw rate and drift) can be obtained instantaneously, i.e. in a single measurement cycle, by combining the two velocity profiles. In contrast to state-of-the art tracking approaches, no model assumptions (e.g. motion model) and time filtering are required. The proposed system has no initialization time and no latency in non-linear maneuvers. This increases the accuracy and reduces the time needed to react to a potential critical situation. The ego-motion can be determined in a single measurement cycle by analyzing the velocity profile of the received stationary detections. An algorithm is developed for a combined estimation using an arbitrary number of radars at any positions. Stationary detections are directly identified by a robust estimator. Significantly, the velocity estimation is highly accurate and the ego-motion is stable even during highly dynamic maneuvers (lateral drift). In addition the method is bias free compared to the currently used odometry (gyroscope, wheel-sensors). By combining the radar-based ego-motion and odometry, the sensor alignment can be obtained online and with high precision. The potential scaling and bias error of the odometry can also be calibrated. The high sampling rate and precision of the gyroscope is thus combined with the radar-based ego-motion estimation, which is free of systematic errors.

Published
2017
Full Text
View/download PDF

27. Dehydrogenative TEMPO-Mediated Formation of Unstable Nitrones: Easy Access to N-Carbamoyl Isoxazolines.

Author

Gini, Andrea, Segler, Marwin, Kellner, Dominik, and Mancheño, Olga García

Subjects
ISOXAZOLINE, CHEMICAL synthesis, NITRONES, RING formation (Chemistry), CARBAMOYL compounds
Abstract

N-carbamoyl nitrones represent an important class of reagents for the synthesis of a variety of natural and biologically active compounds. These compounds are generally converted into valuable 4-isoxazolines upon cyclization reaction with dipolarophiles. However, these types of N-protected nitrones are highly unstable, which limits their synthesis, storage and practical use, enforcing alternative lengthy or elaborated synthetic routes. In this work, a 2,2,6,6-tetramethylpiperidin-1-oxyl (TEMPO)-mediated formal 'dehydrogenation' of N-protected benzyl-, allyl- and alkyl-substituted hydroxylamines followed by in situ trapping of the generated unstable nitrones into N-carbamoyl 4-isoxazolines is presented. A plausible mechanism is also proposed, in which the dipolarophile shows an important assistant role in the generation of the active nitrone intermediate. This simple protocol avoids the problematic isolation of N-carbamoyl protected nitrones, providing new synthetic possibilities in isoxazoline chemistry. [ABSTRACT FROM AUTHOR]

Published
2015
Full Text
View/download PDF

30. Pd(OAc) 2 /Ph 3 P-catalyzed dimerization of isoprene and synthesis of monoterpenic heterocycles.

Author

Kellner D, Weger M, Gini A, and Mancheño OG

Abstract

The palladium-catalyzed dimerization of isoprene is a practical approach of synthesizing monoterpenes. Though several highly selective methods have been reported, most of them still required pressure or costly ligands for attaining the active system and desired selectivity. Herein, we present a simple and economical procedure towards the tail-to-tail dimer using readily available Pd(OAc) 2 and inexpensive triphenylphosphine as ligand. Furthermore, simple screw cap vials are employed, allowing carrying out the reaction at low pressure. In addition, the potential of the dimer as a chemical platform for the preparation of heterocyclic terpenes by subsequent (hetero)-Diels-Alder or [4 + 1]-cycloadditions with nitrenes is also depicted.

Published
2017
Full Text
View/download PDF

31. Dehydrogenative TEMPO-Mediated Formation of Unstable Nitrones: Easy Access to N-Carbamoyl Isoxazolines.

Author

Gini A, Segler M, Kellner D, and Mancheño OG

Abstract

N-carbamoyl nitrones represent an important class of reagents for the synthesis of a variety of natural and biologically active compounds. These compounds are generally converted into valuable 4-isoxazolines upon cyclization reaction with dipolarophiles. However, these types of N-protected nitrones are highly unstable, which limits their synthesis, storage and practical use, enforcing alternative lengthy or elaborated synthetic routes. In this work, a 2,2,6,6-tetramethylpiperidin-1-oxyl (TEMPO)-mediated formal "dehydrogenation" of N-protected benzyl-, allyl- and alkyl-substituted hydroxylamines followed by in situ trapping of the generated unstable nitrones into N-carbamoyl 4-isoxazolines is presented. A plausible mechanism is also proposed, in which the dipolarophile shows an important assistant role in the generation of the active nitrone intermediate. This simple protocol avoids the problematic isolation of N-carbamoyl protected nitrones, providing new synthetic possibilities in isoxazoline chemistry., (© 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.)

Published
2015
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results