Modern vehicles increasingly rely on sensors to monitor their environment and to support driver assistance and safety systems. Most vehicles use a variety of different sensors to improve robustness. A vital part of these is the radar sensor. It provides the vehicle not only with location but also with valuable velocity information from surrounding objects. The increasing usage of radar systems in road traffic also causes problems in terms of mutual interference between different radar sensors. This interference leads to broadband disturbances in the signal which must be mitigated to ensure reliable object detection and object angle estimation. In this article, we compare different variants of convolutional neural networks (CNNs) in their ability to mitigate mutual interference for multiantenna radar data. We analyze the potential of using multiantenna data for real-valued CNN (RVCNN) and complex-valued (CVCNN) models, comparing detection, phase reconstruction, and angle estimation performances. Furthermore, we propose a complex-valued CVCNN (CVCNN) architecture using a modified batch normalization method that omits activation scaling. Our experiments show, that using multiantenna data in combination with CVCNNs can greatly improve detection, phase, as well as angle estimation performance and that activation scaling is detrimental to our CVCNN architecture.

In automotive applications, frequency modulated continuous wave (FMCW) radar is an established technology to determine the distance, velocity and angle of objects in the vicinity of the vehicle. The quality of predictions might be seriously impaired if mutual interference between radar sensors occurs. Previous work processes data from the entire receiver array in parallel to increase interference mitigation quality using neural networks (NNs). However, these architectures do not generalize well across different angles of arrival (AoAs) of interferences and objects. In this paper we introduce fully convolutional neural network (CNN) with rank-three convolutions which is able to transfer learned patterns between different AoAs. Our proposed architecture outperforms previous work while having higher robustness and a lower number of trainable parameters. We evaluate our network on a diverse data set and demonstrate its angle equivariance.

Algorithms for mutual interference mitigation and object parameter estimation are a key enabler for automotive applications of frequency-modulated continuous wave (FMCW) radar. In this paper, we introduce a signal separation method to detect and estimate radar object parameters while jointly estimating and successively canceling the interference signal. The underlying signal model poses a challenge, since both the coherent radar echo and the non-coherent interference influenced by individual multipath propagation channels must be considered. Under certain assumptions, the model is described as a superposition of multipath channels weighted by parametric interference chirp envelopes. Inspired by sparse Bayesian learning (SBL), we employ an augmented probabilistic model that uses a hierarchical Gamma-Gaussian prior model for each multipath channel. Based on this, an iterative inference algorithm is derived using the variational expectation-maximization (EM) methodology. The algorithm is statistically evaluated in terms of object parameter estimation accuracy and robustness, indicating that it is fundamentally capable of achieving the Cramer-Rao lower bound (CRLB) with respect to the accuracy of object estimates and it closely follows the radar performance achieved when no interference is present.

Identifying companies with similar profiles is a core task in finance with a wide range of applications in portfolio construction, asset pricing and risk attribution. When a rigorous definition of similarity is lacking, financial analysts usually resort to 'traditional' industry classifications such as Global Industry Classification System (GICS) which assign a unique category to each company at different levels of granularity. Due to their discrete nature, though, GICS classifications do not allow for ranking companies in terms of similarity. In this paper, we explore the ability of pre-trained and finetuned large language models (LLMs) to learn company embeddings based on the business descriptions reported in SEC filings. We show that we can reproduce GICS classifications using the embeddings as features. We also benchmark these embeddings on various machine learning and financial metrics and conclude that the companies that are similar according to the embeddings are also similar in terms of financial performance metrics including return correlation.

Brake creep groan is a severely annoying noise and vibration phenomenon. Especially on the Asian market, customer feedback about creep groan is common, indicating creep groan’s impact towards the quality impression of a car. Hence, treatment of these stick–slip-related creep groan phenomena is necessary. As numerous design conflicts exist for brake and axle, a complete mitigation of the phenomenon is often not possible. A reduction of creep groan’s annoyance by changing the noise’s level and characteristics is therefore typically aspired. One approach towards this goal could include the usage of psychoacoustics: This work deals with psychoacoustic characteristics of different creep groan classes. Low-frequency groan, high-frequency groan, and transition groan classes are compared regarding loudness, sharpness, roughness, fluctuation strength, and tonality. Standard statistic methods as well as machine learning approaches are applied on signals from vehicle tests and half-axle tests. Test results depict the different characteristics of each creep groan class. By mapping the results to the subjective rating of trained test drivers, the annoyance of different classes is compared. Low-frequency groan, dominated by longitudinal axle vibrations, is found to be least annoying. This low annoyance is best depicted by the psychoacoustic parameters loudness and roughness. Presented results allow an optimization of brake system design to reduce creep groan’s annoyance, leading to higher customer satisfaction and a more goal-oriented treatment of this NVH problem.

In this paper we propose a new method for training neural networks (NNs) for frequency modulated continuous wave (FMCW) radar mutual interference mitigation. Instead of training NNs to regress from interfered to clean radar signals as in previous work, we train NNs directly on object detection maps. We do so by performing a continuous relaxation of the cell-averaging constant false alarm rate (CA-CFAR) peak detector, which is a well-established algorithm for object detection using radar. With this new training objective we are able to increase object detection performance by a large margin. Furthermore, we introduce separable convolution kernels to strongly reduce the number of parameters and computational complexity of convolutional NN architectures for radar applications. We validate our contributions with experiments on real-world measurement data and compare them against signal processing interference mitigation methods.

The perception supply chain (SC1) of the ArchitectECA2030 project investigates failure modes, fault detection, and residual risk in perception systems of electrified, connected, and automated (ECA) vehicles. This accounts for the needs of a reliable understanding of the surrounding environment. The three demonstrators of SC1, described in this paper, address steps of a typical ECA usage cycle: charge - drive - restart charging. The foreign object detection (FOD) demonstrator improves safety within a wireless charging system. The robust physical sensors demonstrator creates a more robust perception by detecting failures within fused and single sensor data. The position enhancement demonstrator improves vehicle localization in areas with reduced GNSS signal coverage. All demonstrators are linked to the challenges that occur during the ECA vehicle usage cycle

Knarzende Bremsen sind unangenehm, für Fahrgäste und Passant:innen. Auch nach vielen Jahren der Forschung ist das charakteristisch niederfrequente, stick-slip-induzierte Bremsenknarzen ein Problem in Bezug auf Kundenrückmeldungen und Qualitätseinbußen bei PKWs. Die knarzfreie Bremse ist durchaus möglich, sei es durch aktive Maßnahmen, wie Piezo-basierte Glättung der Reibkurve, oder durch passive Maßnahmen, wie die Optimierung der Reibcharakteristik mittels Belagrezeptur oder Scheibenbeschichtung. In der Regel wird dieses technisch optimale Ziel allerdings nur selten erreicht. Gründe dafür sind hohe Kosten für aktive Systeme oder Designkonflikte innerhalb der zahlreichen Anforderungen an die Reibmaterialien. Somit müssen Kompromisse eingegangen und ein gewisses „Maß“ an Bremsenknarzen zugelassen werden. Wie dieses „Maß“ bzw. die Lästigkeit durch Knarzgeräusche quantifiziert werden kann, ist Inhalt der vorliegenden Arbeit: Wesentliche Aspekte des Testens, der Erkennung, der Klassifizierung und nicht zuletzt der objektiven Bewertung selbst werden aufgezeigt. Die Ergebnisse dieser Arbeit bilden Grundlagen für einen modernen NVH-Entwicklungsprozess, welcher in der Lage ist, die Lästigkeit des Bremsenknarzens exakt zu quantifizieren und damit eine effektive und effiziente Reduzierung der Knarzgeräusche und -vibrationen ermöglicht.

Categorization of mutual funds or Exchange-Traded-funds (ETFs) have long served the financial analysts to perform peer analysis for various purposes starting from competitor analysis, to quantifying portfolio diversification. The categorization methodology usually relies on fund composition data in the structured format extracted from the Form N-1A. Here, we initiate a study to learn the categorization system directly from the unstructured data as depicted in the forms using natural language processing (NLP). Positing as a multi-class classification problem with the input data being only the investment strategy description as reported in the form and the target variable being the Lipper Global categories, and using various NLP models, we show that the categorization system can indeed be learned with high accuracy. We discuss implications and applications of our findings as well as limitations of existing pre-trained architectures in applying them to learn fund categorization.

Groaning brakes are unpleasant, for passengers and passers-by. Even after many years of research, the characteristic low-frequency, stick-slip-induced brake creep groan is still a problem in terms of customer feedback and quality perception of passenger cars. Groan-free braking is certainly possible, whether through active measures such as piezo-based smoothing of the friction curve, or through passive measures such as the optimization of the friction characteristics by means of pad formulation or disc coating. However, this technically optimal goal is rarely achieved. The reasons for this are high costs for active systems or design conflicts within the numerous requirements placed on the friction materials. Hence, compromises have to be made and a certain “amount” of brake creep groan has to be allowed. How to quantify this “amount” resp. the annoyance due to creep groan noise is the content of the present work: essential aspects of testing, detection, classification and last but not least the objective rating itself are shown. The results of this work form the basis for a modern NVH development process, which is able to precisely quantify the annoyance of brake creep groan and therefore enables an effective and efficient reduction of creep groan noise and vibration.

The critical issue of mutual interference in automotive radar leads to a growing number of processing algorithms for mitigation. Analyses of their performance have been mostly based on either simulations or on a small number of exemplary measurements. In this contribution, a corpus of real-world radar measurements is used, to which strongly varying simulated interferences are added. A statistical analysis framework is then applied to a set of state-of-the-art mitigation algorithms, studying the impact of highly complex object environments on their performance.

Autonomous driving highly depends on capable sensors to perceive the environment and to deliver reliable information to the vehicles' control systems. To increase its robustness, a diversified set of sensors is used, including radar sensors. Radar is a vital contribution of sensory information, providing high resolution range as well as velocity measurements. The increased use of radar sensors in road traffic introduces new challenges. As the so far unregulated frequency band becomes increasingly crowded, radar sensors suffer from mutual interference between multiple radar sensors. This interference must be mitigated in order to ensure a high and consistent detection sensitivity. In this paper, we propose the use of Complex-Valued Convolutional Neural Networks (CVCNNs) to address the issue of mutual interference between radar sensors. We extend previously developed methods to the complex domain in order to process radar data according to its physical characteristics. This not only increases data efficiency, but also improves the conservation of phase information during filtering, which is crucial for further processing, such as angle estimation. Our experiments show, that the use of CVCNNs increases data efficiency, speeds up network training and substantially improves the conservation of phase information during interference removal.

Radar sensors are crucial for environment perception of driver assistance systems as well as autonomous vehicles. With a rising number of radar sensors and the so far unregulated automotive radar frequency band, mutual interference is inevitable and must be dealt with. Algorithms and models operating on radar data are required to run the early processing steps on specialized radar sensor hardware. This specialized hardware typically has strict resource-constraints, i.e. a low memory capacity and low computational power. Convolutional Neural Network (CNN)-based approaches for denoising and interference mitigation yield promising results for radar processing in terms of performance. Regarding resource-constraints, however, CNNs typically exceed the hardware's capacities by far. In this paper we investigate quantization techniques for CNN-based denoising and interference mitigation of radar signals. We analyze the quantization of (i) weights and (ii) activations of different CNN-based model architectures. This quantization results in reduced memory requirements for model storage and during inference. We compare models with fixed and learned bit-widths and contrast two different methodologies for training quantized CNNs, i.e. the straight-through gradient estimator and training distributions over discrete weights. We illustrate the importance of structurally small real-valued base models for quantization and show that learned bit-widths yield the smallest models. We achieve a memory reduction of around 80% compared to the real-valued baseline. Due to practical reasons, however, we recommend the use of 8 bits for weights and activations, which results in models that require only 0.2 megabytes of memory.

The nature of friction within a vehicle’s disk brake system can cause a wide range of different noise phenomena. Especially high-frequency brake squeal was examined during the last decades. Numerous publications treat squeal phenomenology and its mitigation. Increasing shares of electrified powertrains, automatic driving functions such as park assists and further increasing quality demands have now shifted the research interest more and more towards low-frequency phenomena. One of these low-frequency phenomena is creep groan. Defined by its main frequency below 200 Hz, creep groan is characterized by a highly non-linear behavior: Global stick-slip transitions in the disk/pad contacts repeatedly excite the whole brake and axle system. Different bifurcations or even chaotic behavior occur. To ensure good creep groan behavior, defined assessment procedures and rating criteria are necessary. Currently, the German Association of the Automotive Industry recommends a combined rating via the subjective perception of trained test drivers and the objective, A-weighted sound pressure level. This practice could be improved with a more sophisticated objective rating: By considering the human perception, objective and subjective ratings would correlate even better. One possible approach towards an enhanced objective creep groan rating could therefore use psychoacoustic metrics. In 2009, this idea was formulated for the psychoacoustic loudness and the tonality of creep groan by Abdelhamid and Bray. The present work seizes this suggestion and provides additional psychoacoustic evaluations of full-vehicle creep groan signals. Based on measured accelerometer signals, a novel procedure for the psychoacoustic evaluation of structure-borne noise was applied: Optimized FIR filter transfer functions were used to compute equivalent sound pressure signals from the accelerometer data, with the equivalent signals resembling the measured signals but lacking unwanted noise. Both the measured and the simulated signal were then evaluated and compared regarding their psychoacoustic behavior. Results reveal the value of the equivalent sound pressure signal: Whereas loudness and sharpness were found very similar and tonality rather arbitrary for both measured and equivalent sound pressure signal, roughness and fluctuation strength showed strong differences between the signals: Here, only the accelerometer-based, equivalent sound pressure provided easily interpretable characteristics. The proposed method also compared psychoacoustic characteristics for different creep groan bifurcations. Possible applications comprise an enhanced objective rating of low-frequency noise phenomena, the detection and classification of creep groan bifurcations, or the possibility to estimate creep groan cabin noise based on simulative results during early development stages. Therefore, this study provides another step towards silent automotive brake technology.

Radar sensors are crucial for environment perception of driver assistance systems as well as autonomous cars. Key performance factors are a fine range resolution and the possibility to directly measure velocity. With a rising number of radar sensors and the so far unregulated automotive radar frequency band, mutual interference is inevitable and must be dealt with. Sensors must be capable of detecting, or even mitigating the harmful effects of interference, which include a decreased detection sensitivity. In this paper, we evaluate a Convolutional Neural Network (CNN)-based approach for interference mitigation on real-world radar measurements. We combine real measurements with simulated interference in order to create input-output data suitable for training the model. We analyze the performance to model complexity relation on simulated and measurement data, based on an extensive parameter search. Further, a finite sample size performance comparison shows the effectiveness of the model trained on either simulated or real data as well as for transfer learning. A comparative performance analysis with the state of the art emphasizes the potential of CNN-based models for interference mitigation and denoising of realworld measurements, also considering resource constraints of the hardware.

Automotive radars are increasingly used for safety-critical advanced driver assistance systems (ADAS) and autonomous driving applications. Due to unregulated waveforms in the automotive frequency bands, mutual interference between radar-equipped vehicles is becoming a key issue, as it can lead to significant losses of sensor performance. In this work, a novel mutual interference model and a subsequent mitigation method are introduced for chirp sequence radars. The method integrates well into the conventional range-Doppler processing chain, and enables control over the inherent trade-off between interference suppression and signal distortion. Its properties are discussed using extensive simulations within a statistical performance analysis framework. In the investigated scenarios, the method achieved an improvement in detection performance without introducing false alarms.

The number of cars equipped with radar sensors is steadily increasing. Hence, mutual sensor interference will inevitably occur frequently. Interference results in decreased sensing performance, potentially leading to dangerous traffic situations. For the chirp sequence (CS) radars prevalent in automotive applications, different processing algorithms for interference mitigation have been proposed. However, the performances of these algorithms have not typically been evaluated and compared in a systematic way. In this paper we describe a framework based on the statistical evaluation of performance measures, suitable for a general comparison of mutual radar interference mitigation algorithms. The proposed methodology is then applied to analyze selected algorithms based on radar simulations.

Radar sensors are crucial for environment perception of driver assistance systems as well as autonomous cars. Key performance factors are a fine range resolution and the possibility to directly measure velocity. With a rising number of radar sensors and the so far unregulated automotive radar frequency band, mutual interference is inevitable and must be dealt with. Sensors must be capable of detecting, or even mitigating the harmful effects of interference, which include a decreased detection sensitivity. In this paper, we evaluate a Convolutional Neural Network (CNN)based approach for interference mitigation on real-world radar measurements. We combine real measurements with simulated interference in order to create inputoutput data suitable for training the model. A finite sample size performance comparison shows the effectiveness of the model trained on either simulated or real data as well as for transfer learning. A comparative performance analysis with the state of the art emphasizes the potential of CNN-based models for interference mitigation and denoising of real-world measurements, also considering resource constraints of the hardware.

Driver assistance systems as well as autonomous cars have to rely on sensors to perceive their environment. A heterogeneous set of sensors is used to perform this task robustly. Among them, radar sensors are indispensable because of their range resolution and the possibility to directly measure velocity. Since more and more radar sensors are deployed on the streets, mutual interference must be dealt with. In the so far unregulated automotive radar frequency band, a sensor must be capable of detecting, or even mitigating the harmful effects of interference, which include a decreased detection sensitivity. In this paper, we address this issue with Convolutional Neural Networks (CNNs), which are state-of-the-art machine learning tools. We show that the ability of CNNs to find structured information in data while preserving local information enables superior denoising performance. To achieve this, CNN parameters are found using training with simulated data and integrated into the automotive radar signal processing chain. The presented method is compared with the state of the art, highlighting its promising performance. Hence, CNNs can be employed for interference mitigation as an alternative to conventional signal processing methods. Code and pre-trained models are available at https://github.com/johanna-rock/imRICnn.