Refine
Institute
- Fakultät Informatik (597)
- Fakultät Elektro- und Informationstechnik (463)
- THI Business School (412)
Version
- published (858)
Durch die voranschreitende Entwicklung der Technologie im Bereich der
Automobilindustrie können immer mehr digitale Spuren in Fahrzeugen festgestellt werden. Das führt zu einem immer wichtiger werdenden Gebiet, der digitalen Fahrzeugforensik. Dieses beschäftigt sich mit dem Auslesen der Fahrzeugspeicher, die meist proprietäre Dateitypen der Automobilhersteller enthalten. Somit ist der Aufbau dieser Dateien meist unbekannt, was einen Unterschied zur digitalen Forensik darstellt. Hier können durch sogenannte File Carver Dateitypen, anhand bekannter Byte Sequenzen, wie Header oder Footer, erkannt werden. Unbekannte proprietäre Dateien, wie die der Automobilindustrie können somit meist nicht gefunden werden.
Das Ziel dieser Arbeit ist es zu untersuchen, in wieweit klassische File Carver die spezifischen Dateitypen der Automobilbranche erkennen, und ob KI-basierte Ansätze hier möglicherweise einen Vorteil bieten können. Hierzu wird ein synthetischer Datensatz erstellt, um eine Basis mit relevanten Dateitypen zu schaffen.
Die Tests der Softwares auf dem erstellten Datensatz zeigen, dass sich der File Carver Autopsy am Besten für eine Untersuchung in der digitalen Fahrzeugforensik eignet. Jedoch lassen die erzielten Ergebnisse der KI-basierten Methoden auf ein deutliches Entwicklungspotential schließen.
Collaborative perception in automated vehicles leverages the exchange of information between agents, aiming to elevate perception results. Previous camera-based collaborative 3D perception methods typically employ 3D bounding boxes or bird's eye views as representations of the environment. However, these approaches fall short in offering a comprehensive 3D environmental prediction. To bridge this gap, we introduce the first method for collaborative 3D semantic occupancy prediction. Particularly, it improves local 3D semantic occupancy predictions by hybrid fusion of (i) semantic and occupancy task features, and (ii) compressed orthogonal attention features shared between vehicles. Additionally, due to the lack of a collaborative perception dataset designed for semantic occupancy prediction, we augment a current collaborative perception dataset to include 3D collaborative semantic occupancy labels for a more robust evaluation. The experimental findings highlight that: (i) our collaborative semantic occupancy predictions excel above the results from single vehicles by over 30%, and (ii) models anchored on semantic occupancy outpace state-of-the-art collaborative 3D detection techniques in subsequent perception applications, showcasing enhanced accuracy and enriched semantic-awareness in road environments.
Numerous prognostic factors are currently assessed histopathologically in biopsies of canine mast cell tumors to evaluate clinical behavior. In addition, PCR analysis of the c-Kit exon 11 mutational status is often performed to evaluate the potential success of a tyrosine kinase inhibitor therapy. This project aimed at training deep learning models (DLMs) to identify the c-Kit-11 mutational status of MCTs solely based on morphology without additional molecular analysis. HE slides of 195 mutated and 173 non-mutated tumors were stained consecutively in two different laboratories and scanned with three different slide scanners. This resulted in six different datasets (stain-scanner variations) of whole slide images. DLMs were trained with single and mixed datasets and their performances was assessed under scanner and staining domain shifts. The DLMs correctly classified HE slides according to their c-Kit 11 mutation status in, on average, 87% of cases for the best-suited stain-scanner variant. A relevant performance drop could be observed when the stain-scanner combination of the training and test dataset differed. Multi-variant datasets improved the average accuracy but did not reach the maximum accuracy of algorithms trained and tested on the same stain-scanner variant. In summary, DLM-assisted morphological examination of MCTs can predict c-Kit-exon 11 mutational status of MCTs with high accuracy. However, the recognition performance is impeded by a change of scanner or staining protocol. Larger data sets with higher numbers of scans originating from different laboratories and scanners may lead to more robust DLMs to identify c-Kit mutations in HE slides.
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.
The future of automated mobility aims to create significant opportunities for growth and prosperity for the population. An appropriate level of understanding of automated driving, especially in the context of automated shuttle buses, is crucial for improving road safety. Nevertheless, there are many constructs that attempt to predict the factors for people’s perception of automated shuttle buses, but there is no known study to date that has directly examined whether people’s personality traits influences the perception and interaction with such vehicles. This study examines the relationship between personality traits and people’s perception with fully automated shuttle buses. We conducted a study with 40 participants in a virtual CAVE to gain an in-depth understanding of the correlation between the Big Five personality dimensions and the factors influencing the perception of automated shuttle buses. We show that certain personality traits have a main effect on the influencing factors for automated shuttle buses. We found that four of the five dimensions of the Big Five have a significant impact on how people perceive the interaction with an automated shuttle bus. This is particularly seen in the context of Trust, Usability and Acceptance. The results showed that Agreeableness, Conscientiousness and Extraversion had a positive effect on Technology Acceptance, while Neuroticism had a negative effect. Conscientiousness also had a positive effect on the Usability and Agreeableness and Extraversion a positive effect on Trust. Furthermore, it was found that for some personality traits (especially for Agreeableness) the presence of the eHMI has a very high moderate impact on the correlation with the influencing factors. The results are consistent with other studies relating to the Big Five and technology. Suggestions for future research are discussed.
The market for unmanned aerial systems (UASs), also known as unmanned aerial vehicles (UAVs) or drones, is growing rapidly, particularly in the field of Vertical Take Off and Landing (VTOL) UASs. The sensor configuration of a UAS varies depending on the specific use case and type of UAS deployed. Because of this, the different systems have different possibilities to detect and act on a wide range of emergencies. However, all these systems require an appropriate emergency landing method to address potential dangers during UAS flights, regardless of whether the UAS is piloted or flies autonomously. A literature review was conducted to analyze existing research on emergency landings for UASs and identify the necessary requirements for safe emergency landing (SEL) operations. Based on these findings, a novel SEL method was developed. This method presented in this thesis involves semantically segmenting images from a ground-facing camera onboard of the UAS to extract information about its environment. The segmented image is then undistorted to project it onto a ground plane, assuming a flat earth model, restoring metric information. The resulting projection is geo-tagged with the UAS’s current position to create an instance map of the current point in time. By filtering consecutive instance maps using majority voting, a stable map is aggregated, which is then further utilized to detect potential SEL spots by using its inherent semantic information, and for navigation. The SEL method is implemented via the middleware Robot Operating System version 2 (ROS2) using a set of packages responsible for landing spot detection and communication with the flight controller. The development of the SEL implementation included continuous testing in software in the loop (SITL) simulations and real-world test flights, which validated both individual elements of the SEL pipeline and the entire method, resulting in successful demonstrated fully autonomous SEL operations. It can be concluded that the developed method meets all the identified requirements for SEL operations.
In der medizinischen Bildgebung ist die Aufnahme durch Magnetresonanztomographie (MRT) wohl eine der gängigsten Methoden, wenn es um die Informationsgewinnung von Gewebestrukturen geht. Vor allem im Gehirn, wo jeder Eingriff ein hohes gesundheitliches Risiko birgt, bietet das MRT eine sichere Methode, dieses zu analysieren. Als Erweiterung davon kann mittels Diffusions-Bildgebung auch die Bewegung von Protonen im Gewebe gemessen und daraus der Verlauf von Nervenfasern rekonstruiert werden. Diese Methode ist als Traktographie oder auch Fiber Tracking bekannt und kann über verschiedene Algorithmen umgesetzt werden.
In dieser Arbeit soll der Fokus auf Fibertracking mit KI-gestützten Methoden gesetzt werden, um eine Pipeline zu schaffen, die auf beliebige rohe Diffusionsdaten anwendbar ist. Der Aufbau setzt sich dabei aus verschiedenen Schritten zusammen. Zunächst werden die Daten vorverarbeitet, um einen Ground Truth zu erzeugen, der die benötigten Koeffizienten zur Traktographie beinhaltet. Diese können dann im zweiten Schritt von einem Transformer-basierten KI-Modell trainiert werden. Damit soll im letzten Schritt dann aus den geschätzten Werten die eigentliche Traktographie umgesetzt werden. Mit der Verwendung einer gut trainierten KI-Architektur soll so Zeit und Rechenleistung im gesamten Prozess eingespart werden können, da das Modell aufwendige Algorithmen zur Berechnung der Koeffizienten ersetzen kann. Es soll dabei gezeigt, werden wie die komplette Anwendung im Trainingsablauf aussieht, und welche Anpassungen und Optimierungen möglich sind. Die resultierende Code-Pipeline soll den Grundstein für weitere Forschung an Daten von MS-Patienten legen.
Als Betreibergesellschaft eines der wichtigsten Flughäfen Europas muss Flughafen München GmbH stets daran arbeiten, ihre Systeme auf dem technisch aktuellsten Stand zu halten. Aktueller Teil dieser Bemühungen ist die konzernweite Einführung eines auf Microsoft 365 aufbauenden cloudbasierten Standardarbeitsplatzes. Diese Arbeit begleitet dieses Programm und hebt dabei mehrere Aspekte hervor, in denen es ein Vorreiter für die Modernisierung des Konzerns ist. Sie kommt dabei zu dem Schluss, dass in mehrerlei Hinsicht positive Grundlagen für eine weitergehende Entwicklung geschaffen wurden. Vor allem im Bezug auf die Strategie, mit der diese Entwicklung an die Belegschaft kommuniziert wird, findet diese Arbeit jedoch Defizite, die hauptsächlich aufgrund zu spät eingeräumter managementseitiger Aufmerksamkeit für dieses Thema entstanden sind.
Virtual Reality (VR) has emerged as a promising and effective tool for education. The serenity of nature and the underlying Biophilic design theory promise positive effects on learning by influencing attention, well-being and immersion in advanced technology. However, the specifics of how and under what contexts learning is most effective in VR with a focus on the abstraction level remains to be elucidated. This study aims to determine how radically different virtual environments (VE) - natural vs. abstract - design approaches affect learning and attention and the underlying interplay of these two constructs with attention, comfort, presence, and immersion. A mixed-method design with a comparative exploratory approach within subjects was adopted to explore the influence of different VR settings, one being natural and one being abstractly designed without elements. The study emphasizes the crucial role of interaction between presence and attention in VE design. However, while there were no significant quantitative differences regarding this data set, naturalistic VR environments significantly influenced learning effectiveness and attention spans according to qualitative data about the same result. Also, biophilic elements facilitate emotional comfort while considering user comfort and engagement during thematic analysis. It is also evident that quantitative measuring within Immersion and Presence effects reveal slight trends towards natural VE, while qualitative data favored the abstract VE, which raises the question of sensitivity on the MPS Questionnaire and incorporating and elevating the perspective of Acceptance of Virtual Reality Environments to be more sensitive and valid. Our research confirms VR’s potential in education and highlights that traditional classroom elements should blend with VR capabilities (educational VR spaces should augment, not replace, conventional methods). The broader explorative approach revealed potential gaps in VR research, new perspectives on abstraction levels, and the impact on VR learning.
In this thesis, the designing and detailing of a patented hydroelectric energy storage system is documented. Hydropower has been one of the oldest sources of electricity for humankind. This thesis is focused on the idea of manufacturing a hydropower plant to a scale that would be feasible and useful for an average household of 4 people. It also includes the processes and guidelines to manufacture the parts of this energy storage. In current times, renewable sources of energy production are increasing, relevant to earlier years, to reduce the dependence of the world energy needs on fossil fuels. Storing this renewable energy is important and due to limited natural resources like Lithium, Nickel etc. for Battery storage, other viable storage technologies need to be developed such as HEES. Hydro-electric Energy Storage (HEES) is an innovative energy storage concept that utilizes hydropower as energy storage. This paper focuses on the design of the Pelton turbine and the nozzle for a prototype of the HEES. Specifically, the runner, the method used to fix the Buckets on a Pelton runner, and the casing for the runner are analyzed in detail. The HEES nozzle is a critical component responsible for precisely directing high-pressure water jets onto the turbine's rotating buckets.
The study explores the intricacies of Pelton turbine design, including considerations for bucket width, depth, and attachment methods. The importance of nozzle technology for creating ultra-thin water jets and the design of a casing for the high-speed Pelton turbine is also examined. Material selection is a critical aspect, involving stainless steels with varying properties, including corrosion resistance, strength, and weldability. The evaluation and comparison of these materials lead to informed decisions for prototype and final designs. In the context of manufacturing, various methods, such as CNC machining, forming, stamping, forging, and powder metallurgy, are assessed. A detailed analysis highlights the strengths and weaknesses of each approach. For the HEES prototype, a manufacturing method is chosen based on cost-effectiveness and suitability for aluminium, while the final design leans toward stainless steel and methods that ensure strength, precision, and durability. In the ever-evolving landscape of renewable energy, this thesis contributes to the development of a promising Hydro-Energy Storage System, laying the foundation for sustainable and efficient household energy storage solutions.
The thermal management systems on military aircraft have become a bottleneck for
deploying novel technologies that produce high-energy, pulsed heat loads. Multiple
approaches for heat rejection of such heat loads have already been considered, but
no meaningful conclusion can be made about which approach is best suited for this
application. This thesis aims to identify the system architecture that is most suited
for dealing with these high-energy pulsed heat loads. This was done by simulating
a Vapor Compression Cycle and two systems using sensible and latent heat storage
using Simcenter Amesim. The systems were assessed regarding their system mass.
The results show that the system utilizing sensible heat storage reduces the overall
system mass by 67% while using latent heat storage increases system mass by 71%
compared to the simple Vapor Compression Cycle. Based on these findings, a clear
recommendation can be made that future research in this field should focus on
developing architectures using sensible heat storage.
The present study investigates the viability of implementing a micro-scale district heating system (DHS) powered by solar energy in the elevated rural regions of Kyrgyzstan. The performance assessment of such a system is achieved through modeling and simulation in Polysun software along with a conducted parametric study to find out the optimal volume of the storage tank to the area of the collectors based on the solar fraction.
A parametric study resulted in the possible system configuration with a solar fraction of 19.4%, achieved with 2,000 collectors (4,060 m2) and a 400 m3 storage tank. The follow-up economic evaluation considers total lifecycle costs, comparing the solar-assisted DHS with an electric and a coal boiler as the auxiliary heater, and an individual stove-based system. The DHS with a coal-fired boiler resulted in a slightly lower levelized cost of heat (LCOH) at 10.03 €-ct/kWh compared to the electric boiler at 11.77 €-ct/kWh, while an individual home coal-fired system is 1.13 €-ct/kWh.
However, LCOH-based energy affordability analysis revealed that the proposed DHS would burden typical households, where the man works and the woman manages the household, exceeding income by 74% (with electric boiler) and 48% (with coal-fired boiler).
Therefore, the study underscores the need for careful economic consideration in solar thermal projects, especially in low-income areas. To facilitate successful implementation in such regions, securing subsidies or financial support during the project's implementation stage is essential to maintain the affordability of these systems for the local population.
The goal of this thesis is to provide a guide that can be used by software developers that wants to include Rust into an existing C++ application.
After a brief summary of the languages, both the basics of integrating Rust code into C++ applications and the reverse way are presented to set a common base of knowledge. Additionally, a few tools and libraries are introduced, namely bindgen, cbindgen, cxx, and the cc crate. As they are being used in the following guide, the toolsets CMake and CMakeRust are presented and explained in more detail to make it easier for developers to include and use them in their own projects.
The guide is separated into three steps which help to identify the required interfaces, implement them and eventually include everything into the C++ compilation process.
Apart from more advanced types based on pointers, the integration is fairly easy and straight-forward. Using structs and their implemented methods, Rust is able to replicate a class structure that is typical for object-oriented C++. After providing the required binding files, the Rust component can easily be called from C++ code like any other function. While external Rust methods need to be called using a function(Object) syntax, they can be also be converted and used via the standard Object.function() syntax by implementing wrapper classes that translate the calls. These can then be included in more complex C++ architectures like inheritance structures. The usage of the aforementioned advanced types, which are based on pointers, is not covered in this thesis, though the cxx library can be used for those.
The compilation for and usage on embedded targets is also possible with its required effort depending on the target architecture. If the necessary target is not officially supported yet, some additional things like the standard library need to be downloaded and compiled manually.
The thesis comes to the conclusion that considering C++ is not officially supported by the Rust FFI, it works very well with the C ABI and apart from one floating point type, no clear limitations or incompatibilities were identified. For interfaces that only work with basic types and to get a general understanding of the topic, this thesis is a good starting point and guide for a developer who is new to the topic.
Context: Software tools integral to the testing process of automotive software must provide evidence that they are functioning as intended. The safety standard ISO 26262 outlines methods for qualifying these tools; the scope of the thesis lies in Method 1c tool qualification. Ensuring compliance has the potential to create bottlenecks in the tool vendor’s development process.
Objectives: This thesis aims to reduce the qualification effort by integrating automation techniques such as continuous practices while complying with ISO 26262 requirements. Therefore, a concept was developed consisting of widely used components to enable its applicability not only to the tool under investigation but also to related tools.
Methods: An internal study was undertaken to investigate TEST-GUIDE, the software tool targeted to the application of the proposed concept. Regarding development, the chosen components are based on the analysis of acknowledged surveys published in 2023. To explore additional automation degrees feasible in tool qualification, a taxonomy is presented.
Results: The developed concept was successfully applied to one feature of TEST-GUIDE, leading to a tool qualification report that represents the validation results. At the same time, room for optimization is identified concerning the qualification of multiple features.
Conclusions: This thesis presents an automated approach to encountering bottlenecks that occur during tool qualification. Additionally, the taxonomy serves as guidance to derive novel concepts according to a desired automation degree in this domain.
Keywords: ISO 26262, Continuous Practices, Tool Qualification, Automation, Taxonomy
Traffic accidents remain a major global health concern, with millions losing their lives and many more suffering life-altering injuries each year. Especially the delay in emergency services' arrival at the accident scene, caused by various factors, can significantly impact the outcome for the victims. Hence, prompt help from bystanders can save lives and decrease damage to health in the case of an accident. Incorporating innovative systems can help improve response time and provide rapid and reliable access to first aid at the crash scene, ultimately contributing to reducing fatalities and disabilities resulting from road traffic accidents.
This thesis presents the development of a first aider request system that is activated by the eCall, automatically identifying and calling first aiders to the accident scene. The system is designed for usage in vehicles and notifies nearby first aiders via smartphone application or in-vehicle infotainment system. The core of this thesis delves into the design, development, and rigorous evaluation of this system. It explores various technical considerations, system configurations, and task distribution strategies, ensuring optimal system functionality and efficiency. By incorporating an intuitive smartphone application and seamless in-vehicle integration, the system prioritizes accessibility and ease of use for potential first aiders.
The results hold promise for enhancing post-crash response. By harnessing the help of nearby first aiders, the system has the potential to reduce casualties, improve patient outcomes, and ultimately save lives. The findings presented in this thesis demonstrate the first aider request system's valuable contribution to advancing road safety and lifesaving efforts.
Emergent Cooperation from Mutual Acknowledgment Exchange in Multi-Agent Reinforcement Learning
(2022)
Peer incentivization (PI) is a recent approach, where all agents learn to reward or to penalize each other in a distributed fashion which often leads to emergent cooperation. Current PI mechanisms implicitly assume a flawless communication channel in order to exchange rewards. These rewards are directly integrated into the learning process without any chance to respond with feedback. Furthermore, most PI approaches rely on global information which limits scalability and applicability to real-world scenarios, where only local information is accessible. In this paper, we propose Mutual Acknowledgment Token Exchange (MATE), a PI approach defined by a two-phase communication protocol to mutually exchange acknowledgment tokens to shape individual rewards. Each agent evaluates the monotonic improvement of its individual situation in order to accept or reject acknowledgment requests from other agents. MATE is completely decentralized and only requires local communication and information. We evaluate MATE in three social dilemma domains. Our results show that MATE is able to achieve and maintain significantly higher levels of cooperation than previous PI approaches. In addition, we evaluate the robustness of MATE in more realistic scenarios, where agents can defect from the protocol and where communication failures can occur. We also evaluate the sensitivity of MATE w.r.t. the choice of token values.
Die vorliegende Bachelorarbeit dokumentiert die Entwicklung einer Webapplikation zur Optimierung des Bestellprozesses für Arbeitskleidung im FORVIA-Werk Neuburg. Angesichts der ineffizienten und fehleranfälligen papierbasierten Bestellmethode, die das Unternehmen zweimal jährlich durchführt, wurde ein digitalisierter Ansatz entwickelt, um den Prozess zu modernisieren und die Produktivität signifikant zu steigern. Das Hauptziel dieser Arbeit war es, eine hochwertige und benutzerzentrierte Webanwendung zu konzipieren und zu entwickeln, welche die Bedürfnisse der Mitarbeiter berücksichtigt und unabhängig von deren Deutschkenntnissen intuitiv bedienbar ist. Der Entwicklungsprozess umfasste die Anforderungsanalyse und das Design in enger Zusammenarbeit mit den Anforderern und der IT-Leitung, um eine Architektur zu definieren, die sowohl die betrieblichen Abläufe als auch die Anforderungen der Endnutzer abbildet. Die Implementierungsphase fokussierte sich auf die Integration bewährter Programmierpraktiken und die Anwendung von .NET-Technologien, insbesondere unter Verwendung des Entity Frameworks für die Datenverwaltung und Radzen-Blazor für die Benutzeroberfläche. Die resultierende Webanwendung wurde nach der Implementierung in einer Testumgebung geprüft und zeigte eine erfolgreiche Funktionsweise. Die Ergebnisse der Arbeit bestätigen, dass die gesetzten Ziele erreicht wurden. Die Anwendung wird in Kürze in Betrieb genommen und verspricht eine deutliche Verbesserung des Bestellprozesses für Arbeitskleidung.
Diese Arbeit widmet sich dem Thema einer kombinierten Messmethode für Sauerstoff und Kohlendioxid. Zunächst werden mit den optischen, elektrochemischen, resistiven und Lumineszenz-Quenching-basierten Methoden vier geeignete Verfahren aus dem Stand der Technik näher kritisch beleuchtet. Anschließend wird untersucht, ob es möglich ist, einen optischen Sensor für beide Gase aus kommerziell erhältlichen Bauteilen und einem STMF103C8-Microcontroller mit 12bit-A/D-Wandler entwerfen. Nach der Durchführung von Tests kommt die Arbeit zum Schluss, dass die Entwicklung eines Dual-Sensors auf Basis optischer Messprinzipien möglich sein kann, aber hierfür noch weitere Versuchsiterationen mit vorgeschlagenen Verbesserungsansätzen notwendig werden.
Die vorliegende Bachelorarbeit befasst sich mit der Entwicklung eines Ventilatorsystems für das Formula Student Team der Technischen Hochschule Ingolstadt, Schanzer Racing Electric e.V., das den Abtrieb durch erzwungene Anströmung der aerodynamischen Elemente steigern soll. Dieses System soll über verschiedene Geschwindigkeitsbereiche effektiv wirken und die Leistung des Rennwagens verbessern. Mithilfe der Simulationssoftware OpenFOAM wird die Position und der Einfluss unterschiedlicher Volumenströme des Ventilators auf den Abtrieb untersucht, um das Fahrzeug innerhalb einer umfassenden CFD-Simulation zu modellieren. Anschließend wird ein Gesamtkonzept für die Integration des Ventilators in das letzjährige Fahrzeug des Teams, den SRe23, entwickelt, das sowohl elektrische als auch mechanische Aspekte umfasst. Mit einer Rundenzeitsimulation wird schließlich evaluiert, welche Vor- und Nachteile das System mit sich bringt und wie viele zusätzliche Eventpunkte durch den Einsatz des Ventilatorsystems potenziell erreicht werden können, basierend auf einem Vergleich mit den Zeiten vergangener Events.
In scalable traffic analysis for smart city applications and the increasing trend of autonomous driving in the foreseeable future, the implementation of multiple sensor systems for environmental perception will become essential. With the possibility of raw data transmission using 5G communication technology, a centralized approach using sensor data fusion of multiple sensors has been implemented to provide better collaborative environmental perception in infrastructure. Multi-target tracking plays an important role in environmental perception and has attracted enormous interest and effort in the research community resulting in approaches with different benefits and shortcomings. The thesis focuses on the implementation and evaluation of different types of model-based approaches for data association problems, namely Global Nearest Neighbour (GNN), Joint Probabilistic Data Association (JPDA) and Multiple Hypothesis Tracking (MHT) in the context of centralized multi-target tracking. The thesis presents methods for data processing and measurement mapping from sensor coordinates to world coordinates. The thesis also presents the use of homography to determine the position and orientation of the sensors. The implementation of the trackers is conducted in MATLAB and the evaluation of the trackers is performed using GOSPA metrics. The results indicate that GNN achieves optimal performance when targets are spatially distant and is computationally the most cost effective. JPDA demonstrates superior efficacy when targets are in close proximity. Overall, the performance of MHT is compromised by its computational expense.
Die Arbeit beschäftigt sich mit der Frage, wie ein bereits erfolgreicher Lebensmittelhändler in Deutschland ein Geschäftsmodell nach dem „Long Tail“-Prinzip ausgestalten kann. Der „Long Tail“ beschreibt die Strategie, auch weniger populäre Produkte, wie vegane oder glutenfreie Artikel, in einem breiten Sortiment anzubieten, um durch die Kumulierung der Erlöse insgesamt den Umsatz zu erhöhen.
Zur Beantwortung der Forschungsfrage wird die aktuelle Situation des deutschen Lebensmittelhandels betrachtet. Anschließend geht die Arbeit auf die vier zentralen Faktoren des Geschäftsmodells ein: die angebotenen Artikel, die Lebensmittelproduzenten, die potenzielle Kundschaft und die logistische Handhabung von Bestellungen.
Um diese Aspekte zu untersuchen, wurde eine ausführliche Literaturanalyse durchgeführt. Zudem liefern Interviews mit Fachexperten von Edeka Südbayern Einblicke in bestehende Abläufe eines Lebensmittelhändlers.
Die Untersuchung führt zum Ergebnis, dass Konsumenten die Vielfalt des Sortiments als entscheidendes Kriterium bei der Wahl eines Lebensmittelhändlers betrachten. Diese Vielfalt soll durch die Nutzung einer Onlineplattform entstehen. Zudem werden spezifische Konsumentengruppen identifiziert, die besonders von dem Geschäftsmodell profitieren können. Die Arbeit stellt anhand konkreter Zahlen dar, welche Artikelmengen jährlich neu auf den Markt kommen, und erläutert Kriterien anhand derer Lebensmittelhändler eine Auswahl an Produkten für die Plattform treffen können. Es zeigt sich außerdem, dass kleine Lebensmittelproduzenten in besonderer Weise von der Plattform profitieren. Mit dem großen Problem der Distribution kämpfen Lebensmittelhändler heute noch immer. Stationäre Lebensmitteleinzelhändler können das „Long Tail“-Sortiment aufgrund des Umfangs und der Vielfalt nicht vorhalten. Deshalb erfolgt die Zustellung erst nach eingegangener Bestellung aus Lagerung beim Großhändler oder Produzenten. Die Zustellung kann der Lebensmittelhändler schrittweise einführen, zunächst mit dem Click & Collect Modell und schließlich mit dem Endziel der Haustürlieferung. Bei der Ausarbeitung des Geschäftsmodells konnte die Arbeit bestehende Strukturen etablierter Lebensmittelhändler als Vorteil gegenüber jüngeren Akteuren identifizieren.
This bachelor thesis introduces and implements a hybrid software architecture capable of connecting to other instances, forming clusters of nodes to share workload and distribute synthetic environment generation. The architecture supports both peer-to-peer (P2P) and centralized patterns, such as the coordinator-worker model. Nodes can outsource tasks, request services, and share global state through event publishing. A key strength is adaptability to diverse use cases and future technologies through plugin loading. Third-party entities can customize the system with tailored plugins. While advantageous in scenarios with processor bottlenecks, distributing workload onto multiple nodes is not a universal solution. Real-time rendering faces challenges due to communication bottlenecks and network latencies, limiting its applicability. However, it shows promise for more resource-intensive rendering techniques like raytracing, or other inherently distributed applications, like in Monte Carlo simulations.
Magnetic resonance imaging (MRI) has assumed a central role in modern medical imaging thanks to its ability to provide detailed images of the inside of the body. This thesis aims to push the boundaries of conventional MRI by demonstrating two deep learning approaches to reconstruct multparametic correlation spectra. The multi-model approach determines each peak in the spectrum with a separate model, but does not generalise and requires knowledge of the number of spectra being searched for. The residual approach is an iterative process. The strongest signal is determined from the current residual and subtracted from the residual. The spectrum is created using the information obtained. This process is repeated until the residual no longer contains any information. A proof of concept was carried out with the restriction that the number of iterations was determined in advance. This thesis shows that the residual approach works and proves to be promising for future work.
Das Erreichen der Treibhausgasneutralität in Deutschland bis 2045 stellt eine große Herausforderung dar. Die Stromversorgung kann hierzu einen relevanten Beitrag leisten. Daher wird in der vorliegenden Arbeit szenariobezogen die Gewährleistung einer CO2-neutralen Stromversorgung in Deutschland im Verlauf des Jahres 2045 untersucht. Insbesondere ist zu klären, inwieweit dies innerdeutsch bewerkstelligt werden kann bzw. ob grüner Wasserstoff zu importieren ist und wie hoch entsprechend die Elektrolyse- sowie die Wasserstoffkraftwerkskapazität sein müsste. Zur Beantwortung dieser Fragen wurden der Stromverbrauch sowie die einzelnen Stromerzeugungen der erneuerbaren Energien als Haupteinflussfaktoren anhand von Szenariotrichtern analysiert. Die absoluten Jahreswerte wurden sodann in Werte auf Kalenderwochenbasis umgewandelt. Hieraus ließen sich wiederum für das Jahr 2045 drei gesamte Szenarien, welche die Haupteinflussfaktoren beinhalten, konzeptionieren und auswerten. Die Ergebnisse zeigen, dass in zwei der drei Szenarien grüner Wasserstoff in Höhe von 253,02 bzw. 563,38 TWh zu importieren ist, um eine CO2-neutrale Stromversorgung im Verlauf des Jahres 2045 zu gewährleisten. Lediglich in dem dritten Szenario, welches von einem durchschnittlichen Anstieg des Stromverbrauchs sowie dem Erreichen der gesetzlichen Ausbauziele für erneuerbare Energien ausgeht, kann die Stromversorgung eigenständig, ohne Import von Wasserstoff, bewerkstelligt werden. Hierzu ist allerdings eine Wasserstoffspeicherkapazität in Deutschland von mindestens 63 TWh erforderlich. Ferner ergibt die Analyse der drei Szenarien, dass in Deutschland bis 2045 szenariobezogen 5,46 bis 56,72 GW Elektrolyse- und 45,51 bis 80,91 GW Wasserstoffkraftwerkskapazität zu installieren ist.
Energy independence of commercial buildings is a critical aspect of modern construction. In the context of a planned office building project in Ekibastuz, Kazakhstan, located in an extreme climate with unstable coal energy tariffs, it is necessary to determine the energy needs and understand how to cover them. This thesis focuses on determining the level of efficiency in the use of electricity, reducing dependence on local energy suppliers, and integrating alternative sources. The energy assessment was carried out using software tools such as Autodesk Revit and DesignBuilder, which allow the integration of building information modeling (BIM), building energy modeling (BEM) and energy supply modeling (ESM) to optimize the energy design.
The results of commercial projects aimed at achieving nearly zero-energy building (nZEB) standards have informed the development of two energy concepts, both aiming to decrease dependency on district heating networks. These concepts encompass electrical and thermal concepts. In the electrical concept, the air-water heat pump (AWHP) and electric boiler demonstrated better energy efficiency compared to the thermal concept, which utilized a biomass boiler in conjunction with solar thermal collectors (STC) to mitigate the impact of coal-fired electricity. In addition to reducing energy dependence in the air handling unit (AHU), it is proposed to introduce a ground water-to-water heat pump (GWWHP), which is connected to the surface exchanger to capture heat leaks from central heating networks. A comparative analysis of installations without and with heating recovery unit (HRU) shows a fivefold reduction in electricity consumption for heating supply air. Solar technologies, including STC and photovoltaic (PV) panels, are integrated into each concept, but the limited use of STC makes a PV system the preferred option. The decisions made during the modeling process resulted in a fourfold reduction in the office building's average annual consumption compared to government requirements. Selection of energy efficient equipment and modeling led to nZEB classification for both concepts, achieving renewable energy system (RES) penetration rate from 16% to 33% of total consumption.
Economic analysis using the annuity method highlights the cost-effectiveness of rooftop PV systems compared to façade PV systems. In addition, wood pellets and solar heat help reduce costs in the thermal concept. The environmental analysis shows the significant reduction in carbon dioxide emissions achieved by the thermal concept, highlighting the importance of alternative energy sources for sustainable urban development.
With the prevalence of smart technology, machines become more and more connected. One possible field for IoT is the agrarian sector, since environmental factors can be recorded and acted upon independently from humans. This is especially true for agrarian indoor environments, such as stables and greenhouses. One acting device in this is the electromotor, controlling windows and shading, which is why in this thesis, a system of mechanically linked motors is set up as a scenario to evaluate the properties of different wireless technologies. After choosing 802.11ah as the most promising candidate, a hardware test between one transmitting and one receiving device within the protocol is set up and the results evaluated. Following that a simulation of larger scenarios is conducted and the results are compared to the hardware test, as well as used to judge whether the technology is suitable for the scenario.
Fault testing in the production line of automotive traction machines is essential to ensure the desired lifetime. Since repetitive partial discharges (PDs) caused by anomalies in the insulation system lead to premature breakdowns of electrical machines, a reliable PD detection is of great importance. This paper proposes deep learning (DL) methods to improve the discrimination of PD from background noise in comparison with the state-of-the-art amplitude based PD detection in the production line. First, a systematic data extraction and labeling procedure is introduced to obtain correctly labeled datasets from arbitrary PD measurements. In addition, datasets are enhanced with low signal-to-noise ratio PD pulses by applying a special data augmentation approach. 13 different convolutional, recurrent and fully connected neural networks are compared for various time-frequency representations of the input signals. Hyperparameters for input transform, network topology and solver are optimized for all 13 combinations to ensure a fair case study. As a result, the two-dimensional convolutional neural network with continuous wavelet transform achieves the best accuracy of around 99.76% on a test dataset of PD signals originating from previously not utilized test objects. All DL models considered in this comparison outperform the state-of-the-art threshold-based PD classification. Even for PD events with an amplitude close to the noise level, the detection rate is still around 95% for the best network. Furthermore, without applying the proposed data augmentation procedure, the DL models investigated are not able to distinguish small PD pulses from noise.
AbstractOur study emphasizes the evolving nature and increasing relevance of corporate foresight (CF) in a volatile, uncertain, complex, and ambiguous world. This research, therefore, contributes to the foresight literature, concerned with the antecedents and motivators of CF as it investigates how external events that are characterized by great uncertainty influence the CF practice of large companies. On the example of the COVID‐19 pandemic and based on a study of 25 interviews, this empirical research reveals that such events provoke an intensified engagement with the future within companies as implied by a greater resource commitment and interest in foresight. Thereby, CF and especially the development of multiple future scenarios, are considered helpful in countering uncertainty and facilitating responsiveness.
Diagnosis of malignancy in oropharyngeal confocal laser endomicroscopy using GPT 4.0 with vision
(2024)
AbstractGraph data models enable efficient storage, visualization, and analysis of highly interlinked data, by providing the benefits of horizontal scalability and high query performance. Clustering techniques, such as K-means, hierarchical clustering, are highly beneficial tools in data mining and machine learning to find meaningful similarities and differences between data points. Recent developments in graph data models, as well as clustering algorithms for graph data, have shown promising results in image segmentation, gene data analysis, etc. This has been primarily achieved through research and development of algorithms in the field of spectral theory, leading to the conception of spectral clustering algorithms. Spectral clustering algorithms have been one of the most effective in grouping similar data points in graph data models. In this paper, we have compiled 16 spectral clustering algorithms and compared their computational complexities, after an overview of graph data models and graph database models. Furthermore, we provided a broad taxonomy to classify most existing clustering algorithms and discussed the taxonomy in detail.
Development of a high-level System Architecture for a cost-effective Automated Valet Parking System
(2023)
Eignet sich mein Unternehmen für die Veröffentlichung eines Podcasts und wie gehe ich dieses Projekt an? Auf diese Fragen soll jene Thesis Antworten geben. Sie bietet Unternehmen eine Orientierung, indem sie Parameter erarbeitet, die zeigen, welche Bedingungen ein Unternehmen erfüllen muss, damit die Implementierung sinnvoll erscheint. Entscheiden sich Unternehmen dafür, wird ein Leitfaden bereitgestellt, der verdeutlicht, welche Schritte bei der Produktion zu gehen sind und was dabei zu beachten ist.
E-Mobilität im Wandel
(2024)
Die vorliegende Bachelorarbeit widmet sich einer Online-Umfrage mit dem Ziel, mit einer demografischen Analyse die Einflussfaktoren auf die Kaufentscheidung für Elektrofahrzeuge zu untersuchen. Der Fokus liegt nun auf batterieelektrische Fahrzeuge (BEV). Die Umfrage konzentriert sich dabei auf Teilnehmer, die repräsentativ für den deutschen Markt sind.
Die Arbeit beginnt mit einer Einleitung in den theoretischen Rahmen und legt die Grundlagen für die darauf aufbauende Untersuchung durch eine Umfrage fest. Es erfolgt eine ausführliche Darstellung der Problemstellung und die derzeit aktuelle Situation bezüglich des Wandels der Automobillandschaft in Richtung Elektromobilität wird erläutert. Im Zuge der Umfrage werden identifizierte Einflussfaktoren wie Reichweite, Anschaffungskosten, Batterieladeinfrastruktur und einige weitere bewertet, um sie anschließend anhand einer Auswertung und Analyse der demografischen Merkmale genauer zu untersuchen. Dabei werden verschiedene Wohlfühlfaktoren hinsichtlich ihrer Wichtigkeit abgefragt, beschrieben und mittels Grafiken veranschaulicht. Anschließend erfolgt die Analyse der Resultate, woraufhin die Schlüsselbotschaften aus den Ergebnissen abgeleitet und beschrieben werden sollen.
Zum Abschluss der Arbeit wird eine kritische Betrachtung anhand der gewonnenen Erkenntnisse durchgeführt. Insgesamt soll diese Arbeit die aktuelle Situation und die Herausforderungen des Elektromobilitätswandels bezogen auf das Kaufverhalten beschreiben und in einigen Aspekten Politik, Hersteller und Kunden einen Anstoß dafür geben, wie sie sich jeweils eine Möglichkeit für Risiken und Chancen ableiten können.
Diese Bachelorarbeit untersucht die Anwendung von Gamification-Strategien in modernen Recruiting-Prozessen. Der Fokus liegt dabei auf der Analyse spezifischer Fälle, in denen Unternehmen wie Jaguar Land Rover und Deloitte digitale Spiele und Wettbewerbe, wie zum Beispiel „Freedom Firefly“ und L'Oréals „Brandstorm“, in ihre Personalbeschaffungsverfahren integriert haben. Ein zentraler Aspekt der Arbeit ist die Betrachtung der Effektivität und Nachhaltigkeit dieser Strategien. Die Arbeit zeigt auf, wie solche innovativen Ansätze dazu beitragen können, Recruitingprozesse nicht nur effizienter, sondern auch ansprechender für potenzielle Bewerber zu gestalten. Dabei wird auch diskutiert, wie die flüchtige Natur digitaler Innovationen und die damit verbundenen Herausforderungen die akademische Forschung und die langfristige Anwendbarkeit solcher Strategien beeinflussen. Anhand von Praxisbeispielen für Gamificationelemente in Recruitingprozessen soll evaluiert werden, ob der Einsatz solcher Elemente Sinn macht und sich bis zum heutigen Tage etablieren konnte.
Diese Bachelorarbeit untersucht die Einflüsse der aktuellen Trends der Globalisierung, Digitalisierung und Nachhaltigkeitsbestrebungen auf die Struktur und Effizienz internationaler Produktionsnetzwerke in der deutschen Automobilindustrie. Die damit in Verbindung stehenden Herausforderungen und Auswirkungen werden herausgearbeitet und auf Basis dessen werden zukünftige Szenarien abgeleitet.
Die Problemstellung konzentriert sich darauf, wie diese Trends die geografische Verteilung von Produktionsstandorten, die technologische Integration und die strategischen Entscheidungen der Unternehmen beeinflusst haben. Dabei liegt der Fokus insbesondere auf der deutschen Automobilindustrie.
Die Ergebnisse dieser Forschung bestätigen, dass die Globalisierung zu einer internationalen Ausweitung der Produktionsstandorte geführt hat, während die Digitalisierung eine Transformation der Produktionsprozesse und eine zunehmende Vernetzung ermöglicht hat. Gleichzeitig haben die Bestrebungen nach Nachhaltigkeit zu einer Neuausrichtung von Produktionsmethoden und einer verstärkten
Verantwortung für ökologische Aspekte geführt.
Zukünftige Entwicklungen werden voraussichtlich von einer fortschreitenden Digitalisierung, verstärkten Nachhaltigkeitsbemühungen und sich ändernden globalen Handelsbeziehungen geprägt sein. Aufgrund der COVID-19-Pandemie und dem Ukrainekrieg, die jeweils in Produktionsunterbrechungen resultierten, wird für die deutsche Automobilindustrie ein Wandel zur Lokalisierung prognostiziert.
Die vorliegende Bachelorarbeit fokussiert sich auf die Konzeption und Rahmenbedingungen zeitgemäßer Arbeitswelten, insbesondere im Kontext mittelständischer Technologieentwicklung unter Verwendung des Campus-Modells. Die gegenwärtige Situation der Arbeitsrealität wurde analysiert und zwei Unternehmen wurden als Fallstudien herangezogen, die bereits das Campus-Modell erfolgreich implementiert haben. Im Zuge dessen wurden nicht nur Bürostrukturen, sondern auch die bisherige Planungsphase eines neuen Bürogebäudes detailliert erfasst. Die Arbeit schließt mit einer fundierten Empfehlung für die Gestaltung zukünftiger Arbeitswelten auf Basis der gesammelten Erkenntnisse und Erfahrungen.
Nachhaltige Verpackungen sind in der Lebensmittelindustrie ein wichtiges Thema,
da sie sowohl ökologische als auch ökonomische Vorteile bieten können. Diese
Bachelorarbeit untersucht, wie Unternehmen durch die strategische Anpassung
des Marketing-Mix und die Integration der Blue Ocean Strategy ihre
Marktakzeptanz und Wettbewerbsfähigkeit verbessern können. Sie verwendet eine
Kombination aus qualitativen und quantitativen Forschungsmethoden, darunter ein
Experteninterview, Literaturrecherche und eine Datenanalyse. Diese Studie zeigt,
dass Nachhaltigkeit umfassend in alle Aspekte des Marketing-Mix integriert werden
muss und innovative Ansätze zur Differenzierung und Marktdurchdringung
entscheidend sind. Somit leistet die vorliegende Arbeit einen Beitrag zur
Forschungsliteratur über nachhaltige Verpackungen, indem sie eine innovative
Strategie zur Differenzierung und Marktdurchdringung vorstellt. Die Arbeit bietet
praktische Empfehlungen für Unternehmen, die sich in einem umweltbewussten
Markt behaupten wollen.
The continuous introduction of new services in modern automotive systems increases their digital capabilities and growing significance for digital forensic investigations. Those investigations are necessary to reconstruct events of potential crimes, generate security learnings, or resolve insurance cases due to the increased attack surface resulting from the increasing digital capabilities of modern vehicles. Consequently, the acquisition of vehicle and vehicle ecosystem components, as well as analysis of the collected data, has become pivotal for investigators seeking to reconstruct events effectively. This dissertation proposes a novel approach to leverage a multi-layered model utilizing ontologies in automotive digital forensics investigations, facilitating more efficient and effective data analysis by combining information from diverse sources. By harnessing ontologies to reuse knowledge collected from past investigations, investigators can seamlessly combine data from multiple sources, unveil hidden knowledge, and establish connections that might have otherwise remained obscure, thereby aiding in the resolution of crimes and apprehending criminals. A multi-layered model is presented to showcase increased effectiveness achieved through knowledge reuse. The model comprises a query layer that serves as an interface between investigators and the knowledge layer, allowing automotive digital forensics stakeholders to pose forensic questions and hypotheses via SPARQL queries. These queries extract data from the knowledge layer, which consists of two ontologies developed for this study: the general vehicle ontology, offering comprehensive knowledge on modern vehicles, and the vehicle type ontology, containing vehicle-specific knowledge, such as information from a Tesla Autopilot. The knowledge is generated through the operational layer, which introduces digital forensics tools generating results in an ontology-ready format. Two real-world investigations were undertaken to validate the effectiveness of the proposed approach: a thorough examination of the Tesla Autopilot system and an in-depth inquiry involving a GM Airbag control module. Both investigations involved crucial evidence items highly relevant for event reconstruction. The results demonstrate a significant increase in the effectiveness of automotive digital forensics investigations using the multi-layered model. This improvement is validated through three evaluation criteria and twelve metrics to show the application proximity of the multi-layered model.
Knowledge Tracing (KT) aims to predict the future performance of students by tracking the development of their knowledge states. Despite all the recent progress made in this field, the application of KT models in education systems is still restricted from the data perspectives: 1) limited access to real life data due to data protection concerns, 2) lack of diversity in public datasets, 3) noises in benchmark datasets such as duplicate records. To resolve these problems, we simulated student data with three statistical strategies based on public datasets and tested their performance on two KT baselines. While we observe only minor performance improvement with additional synthetic data, our work shows that using only synthetic data for training can lead to similar performance as real data.
Levels of Driving Automation
(2023)