Refine
Year of publication
Document Type
- Conference Paper (962)
- Article (736)
- Bachelor Thesis (470)
- Master's Thesis (284)
- Part of a Compilation / Book Chapter (158)
- Publication by THI (82)
- Book (77)
- Preprint (56)
- Report (38)
- Part of Periodical (9)
Language
- English (1974)
- German (902)
- Portuguese (2)
- Czech (1)
Institute
- Fakultät Informatik (556)
- Fakultät Elektro- und Informationstechnik (437)
- THI Business School (376)
- Fakultät Maschinenbau (311)
- Fakultät Elektrotechnik und Informatik (245)
- Human-Computer Interaction Group (HCIG) (207)
- Institut für Innovative Mobilität (IIMo) (198)
- Fakultät Wirtschaftsingenieurwesen (193)
- AImotion Bavaria (189)
- CARISSMA (189)
Review
- peer-review (1211)
- nein (261)
- ja (209)
- editorial review (64)
Version
- published (810)
Cybersecurity as a Service
(2023)
Towards to Battery Digital Passport: Reviewing Regulations and Standards for Second-Life Batteries
(2024)
Greenhouse gas emissions from transportation harm the environment. In response to these environmental concerns, numerous countries encourage the adoption of electric vehicles (EVs) as a more environmentally friendly option than traditional gasoline-powered vehicles. Advances in battery technology have made batteries an alternative solution for energy storage in stationary applications and for electric mobility. Reduced lithium-ion batteries (LIBs) production costs due to economies of scale, electrode material and cell design developments, and manufacturing process improvements have driven this success. This trend is expected to increase the number of LIBs on the market that may be discarded in the environment at the end of their useful life if more sustainable alternatives are not technologically mature. This coming environmental concern can be mitigated by collecting wasted EV batteries, reconfiguring them, and reusing them for applications with less stringent weight, performance, and size requirements. This method would extend battery life and reduce environmental effects. The present work investigates the main regulatory structures of the second-life battery industry that require rules, technical standards, and laws. To achieve this objective, a systematic review was carried out following a strict protocol that includes identifying relevant studies, extracting data and information, evaluating, and summarizing information. This paper explains the primary rules and technical standards governing the second-life battery business. The findings highlight the need for universities, research institutions, and government agencies to evaluate the second-life battery industry objectively. This would enable the creation of new technological regulations and laws for this burgeoning industry.
Multidimensional Magnetic Resonance Imaging (MRI) is a versatile tool for microstructure mapping. We use a diffusion weighted inversion-recovery spin echo (DW-IR-SE) sequence with spiral readouts at ultra-strong gradients to acquire a rich diffusion-relaxation data set with sensitivity to myelin water. We reconstruct 1D and 2D spectra with a two-step convex optimization approach and investigate a variety of multidimensional MRI methods, including 1D multi-component relaxometry, 1D multi-component diffusometry, 2D relaxation correlation imaging, and 2D diffusion-relaxation correlation spectroscopic imaging (DR-CSI), in terms of their potential to quantify tissue microstructure, including the myelin water fraction (MWF). We observe a distinct spectral peak that we attribute to myelin water in multi-component T1 relaxometry, T1-T2 correlation, T1-D correlation, and T2-D correlation imaging. Due to lower achievable echo times compared to diffusometry, MWF maps from relaxometry have higher quality. While 1D multi-component T1 data allows much faster myelin mapping, 2D approaches could offer unique insights into tissue microstructure and especially myelin diffusion.
Removing the bias and variance of multicentre data has always been a challenge in large scale digital healthcare studies, which requires the ability to integrate clinical features extracted from data acquired by different scanners and protocols to improve stability and robustness. Previous studies have described various computational approaches to fuse single modality multicentre datasets. However, these surveys rarely focused on evaluation metrics and lacked a checklist for computational data harmonisation studies. In this systematic review, we summarise the computational data harmonisation approaches for multi-modality data in the digital healthcare field, including harmonisation strategies and evaluation metrics based on different theories. In addition, a comprehensive checklist that summarises common practices for data harmonisation studies is proposed to guide researchers to report their research findings more effectively. Last but not least, flowcharts presenting possible ways for methodology and metric selection are proposed and the limitations of different methods have been surveyed for future research.
Die Lunge des Menschen ist als eines der Hauptorgane für die Sauerstoffversorgung, die Sauerstoffaufnahme und die anschließende Weitergabe an das Blut im menschlichen Körper zuständig. Eine Erkrankung dieses Organs kann erhebliche gesundheitliche Folgen haben. In der Pneumologie ist man daher bestrebt, solche Krankheiten frühzeitig zu erkennen und effektiv zu behandeln. Das Ziel der vorliegenden Arbeit ist es, anhand von verschiedenen Werten des Blutes, mit Fokus auf Immunoglobulin G und Daten über bisherige Erkrankungen die Verläufe von Lungenkrankheiten vorhersagen zu können und die Behandlung dementsprechend anzupassen.
With the rise of autonomous driving technology, concerns over the safety of AI-based systems have increased, particularly regarding their effect on human life and infrastructure. The complexity and unpredictable nature of AI-enabled software functions have disrupted traditional safety evaluation methods. In such a scenario, the quality control methods used at the manufacturing environments of sensors used for deploying AI-based perception functions need to be coupled with the downstream functions. This study focuses on investigating the impact of a specific production-related defect called blemish, which is caused by debris on image sensors of camera modules. The research aims to answer two fundamental questions: 1) How can blemish artifacts be mathematically modeled, and what is the process for developing a suitable dataset that incorporates this defect? 2) How can the influence of blemish artifacts on performance of object detection models be traced back to define quality requirements of the camera modules? This work takes a novel perspective by shifting the focus from improving model resilience to improving hardware quality camera modules by studying the effects of manufacturing defects on downstream functions. This research seeks to establish a framework to derive quality targets for front camera module manufacturing which can be extended to other sensors for similar applications.
Federated learning allows for cooperative training among distributed clients by sharing their locally learned model parameters, such as weights or gradients. However, as model size increases, the communication bandwidth required for deployment in wireless networks becomes a bottleneck. To address this, we propose a residual-based federated learning framework (ResFed) that transmits residuals instead of gradients or weights in networks. By predicting model updates at both clients and the server, residuals are calculated as the difference between updated and predicted models and contain more dense information than weights or gradients. We find that the residuals are less sensitive to an increasing compression ratio than other parameters, and hence use lossy compression techniques on residuals to improve communication efficiency for training in federated settings. With the same compression ratio, ResFed outperforms current methods (weight- or gradient-based federated learning) by over 1.4× on federated data sets, including MNIST, FashionMNIST, SVHN, CIFAR-10, CIFAR-100, and FEMNIST, in client-to-server communication, and can also be applied to reduce communication costs for server-to-client communication.
Automated Valet Parking Systems (AVPS) relieve the driver of the entire parking process. Many of the systems known today rely on a combination of automotive sensors with sensors of the infrastructure. For this purpose, parking facilities are equipped with comprehensive sensor technology to support the vehicles in environment sensing and route planning. This approach is comparatively expensive which is why many parking operators don’t provide that technology to their customers. This paper proposes a lean AVPS system architecture that requires minimal effort to adapt the infrastructure. At the same time, state-of-the-art vehicle technology is used to make AVPS more profitable overall. At the beginning, an overview will be given describing the state of the art of AVPS. Subsequently, requirements for the AVPS will be elaborated, whereby the system can be designed and implemented in the following. Finally, the presentation of simulation results shows that one doesn’t have to extend the infrastructure with sensors to develop a safe and reliable AVPS.
If we, as user experience designers, want to create truly user-centered experiences, we need to understand not only what people want, but what they need and how they behave. Human behavior, however, also results from the choices people make every day. The same applies to the development and design of automated transportation. To achieve and maintain these goals, we need to accept that human decisions are not only made on a rational level, but rather on the basis of mental shortcuts, which may be efficient, but can also lead to cognitive biases and thus to sub-optimal results.
This master’s thesis specifically sheds light on four cognitive biases - truthiness, illusion of control, action bias and automation bias -, their existence in automated driving, as well as their relationship to each other and possible influencing factors. As part of a preliminary study (n=117), which was implemented as an online questionnaire, the existence of the cognitive biases in focus of this work is examined in more detail and initial findings of visualizations and explanations as an influencing factor are collected. In the further course, a user study (n=34) was completed to build on the knowledge gained and examine the accuracy of the mental model of the occupants of an automated ride and their individual personality traits as further influential parameters. This work should serve as a cornerstone to include cognitive biases and their influencing forces in the development of automated vehicles in the future and to use the knowledge gathered to integrate psychological aspects that affect every human being into user-centered humanmachine interactions.
The fatigue behavior of thin electrical steel sheets under cyclic loading is investigated in dependence on the edge surface. Therefore, four different edge conditions are compared, whereas the edge is either laser cut, shear cut, wire cut, or polished. Strain- and stress-controlled fatigue tests are performed to determine S-N curves in the low cycle regime as well as in the high cycle regime. Microstructural data is collected by non-contacting (optical) Profilometry, Nanoindentation, X-Ray Diffraction, and Electron Backscatter Diffraction to understand the differences in fatigue life by considering surface roughness, residual stresses, hardness, and microstructure. Shear cut specimens achieve the lowest fatigue life, while the other edge conditions reach relatively similar values in the LCF regime. Crack initiation is mainly intergranular in the case of defect-free edges. This tendency has a considerable influence on the observed fatigue behavior.
The impact of technology on employment and consequently on skills is becoming increasingly apparent, particularly in the use of artificial intelligence.
The aim of this bachelor thesis can therefore be divided into two research questions: On the one hand, the aim is to analyze what impact the use of artificial intelligence will have on the labor market. The second question is concerned with the impact artificial intelligence will have on the labor market in terms of the skills required.
In order to provide a profound answer to the research questions, a wide range of literature was used, including books, research studies and use cases. Based on the research questions, the thesis addresses the worldwide AI driven labor market, identifying both the displacement of human workers and the simultaneous opportunities for new occupations and economic growth. Moreover, a variety of necessary skills are highlighted that are essential for overcoming the dynamic challenges on the labor market posed by AI. Finally, the subsequent analysis of the financial sector and public administration serves to illustrate and clarify the relevance of these skills. Therefore, both knowledge of artificial intelligence and existing human skills are important. Depending on the occupation and task area, the focus of skills is either on the direct application of artificial intelligence or on developing an awareness and understanding of it. It also shows that a wide range of actors, including companies and educational institutions, have a key role to play in the process of adapting workers' skills.
Environment perception using camera, radar, and/or lidar sensors has significantly improved in the last few years because of deep learning-based methods. However, a large group of these methods fall into the category of supervised learning, which requires a considerable amount of annotated data. Due to uncertainties in multi-sensor data, automating the data labeling process is extremely challenging; hence,
it is performed manually to a large extent. Even though full automation of such a process is difficult, semiautomation can be a significant step to ease this process. However, the available work in this regard is still very limited; hence, in this paper, a novel semi-automatic annotation methodology is developed for labeling RGB camera images and 3D automotive radar point cloud data using a smart infrastructure-based sensor setup. This paper also describes a new method for 3D radar background subtraction to remove clutter and a new object category, GROUP, for radar-based object detection for closely located vulnerable road users. To validate the work, a dataset named INFRA-3DRC is created using this methodology, where 75% of the
labels are automatically generated. In addition, a radar cluster classifier and an image classifier are developed, trained, and tested on this dataset, achieving accuracy of 98.26% and 94.86%, respectively. The dataset and Python scripts are available at https://fraunhoferivi.github.io/INFRA-3DRC-Dataset/.
Crash-Prone Fault Combination Identification for Over-Actuated Vehicles During Evasive Maneuvers
(2024)
Throughout a vehicle’s lifecycle, systems may fail during operation, requiring effective fault management by the vehicle controller. Various system faults affect vehicle handling differently. Additionally, vehicle velocity and road friction directly impact handling and stability. Thus, it is essential to investigate relevant factors, such as actuator faults, vehicle velocity, road friction, and their combinations, before developing a fault-tolerant controller to mitigate potential critical situations. Our work thus focuses on identifying faults and fault combinations that might lead to crashes for over-actuated vehicles during evasive maneuvers and those impacting comfort parameters. We employ a state-of-the-art vehicle controller optimized for evasive lane changes for over-actuated vehicles. The driving scenario encompasses critical conditions defined in ISO 26262 with ASIL-D, including velocities up to 130 km/h and requiring steering away from obstacles. Failure Mode and Effects Analysis, Design of Experiments, and statistical tools are used to determine fault combinations most likely to lead to crashes during evasive maneuvers. Our results indicate that the vehicle controller successfully handled the maneuver in over 53% of investigated cases, reaching up to 75.1% on dry surfaces. Road friction emerges as the most critical parameter for collision avoidance and comfort. Brake faults exhibit a higher influence on vehicle handling than other actuator faults, while single motor faults do not significantly impact vehicle parameters. Regarding two-factor interactions, brake actuators dominate, followed by steering and motor. These findings provide valuable insights for developing fault-tolerant controllers for over-actuated vehicles, guiding decisions on addressing specific faults to enhance safety and comfort parameters.
Accurate and comprehensive semantic segmentation of Bird's Eye View (BEV) is essential for ensuring safe and proactive navigation in autonomous driving. Although cooperative perception has exceeded the detection capabilities of single-agent systems, prevalent camera-based algorithms in cooperative perception neglect valuable information derived from historical observations. This limitation becomes critical during sensor failures or communication issues as cooperative perception reverts to single-agent perception, leading to degraded performance and incomplete BEV segmentation maps. This paper introduces TempCoBEV, a temporal module designed to incorporate historical cues into current observations, thereby improving the quality and reliability of BEV map segmentations. We propose an importance-guided attention architecture to effectively integrate temporal information that prioritizes relevant properties for BEV map segmentation. TempCoBEV is an independent temporal module that seamlessly integrates into state-of-the-art camera-based cooperative perception models. We demonstrate through extensive experiments on the OPV2V dataset that TempCoBEV performs better than non-temporal models in predicting current and future BEV map segmentations, particularly in scenarios involving communication failures. We show the efficacy of TempCoBEV and its capability to integrate historical cues into the current BEV map, improving predictions under optimal communication conditions by up to 2% and under communication failures by up to 19%. The code will be published on GitHub.
Compression of end-to-end non-autoregressive image-to-speech system for low-resourced devices
(2023)
Dogs-of-the-DAX
(2024)
Diese Arbeit überträgt die populäre, US-amerikanische „Dogs-of-the-Dow“-Anlagestrategie auf den deutschen Markt. Sie beantwortet die Frage, ob ein Aktienportfolio basierend auf DAX-Dividendenstrategien von 2013 bis 2023 eine bessere Performance im Vergleich zum DAX hätte erzielen können. Für eine umfassende Bewertung wird zusätzlich eine Performanceanalyse für krisenhafte Zeiträume anhand der COVID-19-Pandemie und des Ukraine-Kriegs durchgeführt.
Repräsentativ für die DAX-Dividendenstrategien stehen dabei das „Top-10“-, „Low-5“-, und „Low-1“-Portfolio sowie der Dividendenindex DivDAX. Das „Top-10“-Portfolio setzt sich aus den zehn dividendenstärksten Aktien des deutschen Aktienindex DAX zusammen. Daneben sind im „Low-5“- bzw. „Low-1“-Portfolio die fünf bzw. eine Aktie der „Top-10“ mit dem niedrigsten Aktienkurs allokiert. Alle Aktien werden gleichgewichtet.
Mithilfe der jährlichen Bestimmung der diskreten Rendite, annualisierten Volatilität und Sharpe Ratio wird die Performance der DAX-Dividendenstrategien im Vergleich zum DAX sowie DAX Equal Weight als Benchmarks analysiert.
Obwohl die „Top-10“ in fünf der elf Jahre eine höhere diskrete Rendite als der DAX erreichten, lag ihre durchschnittlichen Sharpe Ratio mit 0,57 um 0,02 unter der des DAX (0,59). Auch die „Low-5“ erzielten in sechs der elf Jahre eine Überrendite, waren jedoch hinsichtlich ihrer Gesamtperformance um 0,03 unterhalb des DAX positioniert.
Demgegenüber erwies sich das sehr risikoreiche „Low-1“-Portfolio mit einer durchschnittlichen Sharpe Ratio von 0,75 als überlegen gegenüber dem DAX.
Überraschenderweise war das Rendite-Risiko-Profil des DivDAX erfolgreicher als das der „Top-10“ und „Low-5“.
Sämtliche DAX-Dividendenstrategien waren dem DAX Equal Weight risikoadjustiert überlegen, was die Wertpapiergewichtung als Erfolgsfaktor des DAX vermuten lässt.
Customer Perceived Value is at the heart of marketing and business. There are many approaches, but inconsistencies exist in understanding the nature, characteristics and conceptualization of this phenomenon. Therefore, a coherent concept is required. This thesis is an endeavor to go beyond the established terminology to enrich the discipline. It provides an interdisciplinary investigation that combines philosophical, psychological, physical and neuroscience perspectives to understand the nature of Customer Perceived Value and structure knowledge using a new approach. Based on a synthesis of theories, a new metaphorical concept was developed that considers the nature of customer perceived value from the perspective of world-forming concepts of information and energy and their interactions. The uniqueness of the concept lies in its ability to integrate and explain all qualities of Customer Perceived Value and amalgamate all previous conceptions in this field. This new concept provides a practical tool for managing value from a strategic marketing perspective. The new approach also represents the basis for a new Customer Value typology. Not the primary but equally significant goal of this thesis is the popularization of the following theories: Simonov’s Need-Informational theory of emotions, Ostwald's Energetics, Rupert Sheldrake’s Theory of Morphic Fields and Kurt Lewin's Field Theory. This concept, developed based on the insights provided by these theories, contributes to marketing science.
A simulation study to analyse the impact of V2X communication on the emergency vehicle response time
(2024)
Focused on the crucial intersection of urban mobility and public safety, the thesis investigates the practical implications of integrating V2X technology for the streets of Ingolstadt. This study employs simulation techniques to mimic real-world situations, evaluating the tangible benefits of V2X-enabled vehicles in speeding up emergency arrival time.
By connecting theory with simulations, this work aims to guide policymaking and technology development dept. for improved emergency response systems, for the roads of Ingolstadt.