TU Wien

reposiTUm
Not a member yet
    256869 research outputs found

    Bionik - Lernen von der Natur für nachhaltige Technologien

    No full text
    Bionik nutzt die belebte Natur als Ideengeberin für Technologien: Organismen optimieren nicht „maximal“, sondern „gut genug“ – robust, ressourcenschonend und eingebettet in Kreisläufe. Der Vortrag zeigt, wie sich Materialien, Strukturen und Prozesse aus der Natur in nachhaltige technische Ansätze übertragen lassen – mit Fokus auf biobasierte, biologisch abbaubare Werkstoffe und funktionale Strukturen in diesen Materialien. Im Zentrum steht das „Good-Enough“-Prinzip als bionischer Zugang zur Nachhaltigkeit: Produkte und Funktionen sollen reduzierbar, reparierbar, wiederverwendbar oder am Lebensende rückführbar sein (z. B. als Nährstoff/Dünger oder als Ausgangsstoff für weitere Wertschöpfung). Beispiele reichen von Adhäsion und kontrollierter Benetzbarkeit über wasserabweisende bzw. wasserleitende Oberflächen bis zu Strukturfarben, passiver Strahlungskühlung und antibakteriellen Oberflächen – inspiriert u. a. von Sahara-Wüstenameisen und Zikaden. Dabei wird deutlich, wie „funktionale Hierarchie“ in Keratin-, Chitin- und Cellulose-basierten Strukturen technische Lösungen ermöglicht, die zugleich material- und systemisch nachhaltig gedacht sind

    An improved reliability factor for quantitative low-energy electron diffraction

    No full text
    Quantitative low-energy electron diffraction (LEED I(V) or LEED I(E)), which evaluates the dif- fraction intensities I as a function of the electron energy, is a versatile technique for the study of surface structures. The technique is based on optimizing the agreement between experimental and calculated intensities. Today, the most commonly used measure of agreement is Pendry’s R factor RP. While RP has many advantages it also has severe shortcomings, as it is a noisy target function for optimization and very sensitive to small offsets of the intensity. Furthermore, RP = 0, which is meant to imply perfect agreement between two I(E) curves, can also be achieved by qualitatively very different curves. We present a modified R factor RS, which can be used as a direct replacement for RP, but avoids these shortcomings. We also demonstrate that RS is as good as RP or better in steering the optimization to the correct result in the case of imperfections in the experimental data, while another common R factor, RZJ (suggested by Zanazzi and Jona) is worse in this respect

    AFM & STM Fundamentals

    No full text

    Systematic Literature Review: Fair By Design

    No full text
    The rapid adoption of Artificial Intelligence has raised concerns about potential bias,discrimination, and ethical use of social markers. This systematic literature review(SLR) examines the principles and guidelines for distinguishing between fair and unfair discrimination in AI systems, focusing on their use of social markers such as race, gender,age, disability, socioeconomic status, intersectionality, and many more. Guided by research questions, the review investigates the conditions that justify the inclusion of social markers, the implementation of these distinctions in recruitment frameworks, andthe current practices addressing fairness and bias in AI systems. A comprehensive searchand selection process included 91 articles in English and publications from 2018 to 2024.These sources were analyzed to uncover themes related to data bias, proxy discrimination,intersectionality, explainability, and the role of regulatory frameworks. Our analysis reveals that the ethical use of social markers is contingent on transparent, fairness-driven applications designed to mitigate systemic inequities and improve inclusivity. However,our study also points out major hazards, including opaque decision-making procedures,inadequate responsibility, and growing historical prejudices. This study emphasizes the need to include substantial fairness criteria, governance structures, and stakeholderviewpoints in artificial intelligence evolution. This research contributes to the fieldby providing actionable insights into designing AI systems that align with ethical and current legal standards for fairness. It highlights the need for intersectional approaches and continuous auditing to address the complexities of fairness and discrimination in automated decision-making, particularly in AI recruitment contexts. The findings serveas a foundation for future research and development in responsible AI

    Optimierung des Fahrverhaltens von Fahrzeugen durch Überaktuierung

    No full text
    Menschliches Fehlverhalten ist die Hauptursache für Verkehrsunfälle und resultiert häufig aus Fehleinschätzungen der Fahrsituation oder der Fahrzeugfähigkeiten. Fortschritte in der Fahrzeugaktuierung sowie cloud-vernetzte Fahrzeuge bieten vielversprechende Ansätze, um solche Fehler zu mindern, indem sie leistungsfähigere und intelligentere Fahrdynamikregelungen ermöglichen. Diese Dissertation untersucht, wie Überaktuierung das Fahrzeughandling beeinflusst, mit Auswirkungen auf Regler für das autonome Fahren, und wie vorausschauende Informationen über die Fahrumgebung, bereitgestellt durch Fahrzeugvernetzung, proaktive Fahrdynamikregelungsstrategien ermöglichen. Eine optimierungsbasierte Methode wird vorgestellt, die das unter stationären Bedingungen fahrdynamisch machbare Handling unter Berücksichtigung von Aktuatorkonfigurationen und deren Grenzen quantifiziert. Dadurch wird das erreichbare Fahrzeughandling durch Lenk- und Schwimmwinkelcharakteristiken definiert. Innerhalb dieser Bereiche werden Hinterradlenkung (RWS), Torque Vectoring (TV) mittels Einzelradmotoren sowie TV+RWS hinsichtlich ihres Einflusses auf den Fahrzeugleistungsbedarf, die Fahrstabilität, die Reibungspotenzialausnutzung und damit die maximale Normalbeschleunigung bewertet. Die Ergebnisse zeigen, dass die drei Aktuatorkonfigurationen die Charakteristiken wesentlich beeinflussen können, jedoch nur TV und TV+RWS eine gleichzeitige, unabhängige Beeinflussung beider Charakteristiken ermöglichen, wobei TV+RWS ein lineares Fahrzeughandling bis zur maximalen Normalbeschleunigung ermöglicht. Basierend auf dem mit TV+RWS machbaren weitgehend linearen Fahrezughandlings wird eine hierarchische Regelarchitektur für die autonome Pfadfolge vorgeschlagen. Durch das Aufzwingen eines linearen Fahrzeugverhaltens mittels TV+RWS wird ein vereinfachtes Prädiktionsmodell für die modellprädiktive Regelung möglich. Dies ermöglicht längere Prädiktionshorizonte und höhere Recheneffizienz, da Reifennichtlinearitäten und Überaktuierung im Prädiktionsmodell vermieden werden. Analysen und Simulationen zeigen, dass die Architektur nicht nur die Fahrzeugstabilität unter kritischen Bedingungen verbessert, sondern auch den Pfadfolgefehler gegenüber einem Basisfahrzeug reduziert. Im Kontext intelligenter Fahrzeugregelung wird das Potenzial eines cloudbasierten digitalen Zwillings untersucht, der Fahrzeug und Umgebung abbildet, zur Verbesserung der Fahrzeugdynamikregelung durch verfeinerte Fahrzeugmodelle, proaktive Regelungsstrategien und personalisierte Fahrdynamikregelungen in kritischen Fahrsituationen. Neben der theoretischen Diskussion wird ein Anwendungsfall zur Fahrzeugstabilität experimentell untersucht. Die Experimente zeigen, dass adaptive Geschwindigkeitsregelung und TV mithilfe von Informationen des digitalen Zwillings proaktiv die Fahrzeuggeschwindigkeit und die Antriebsmomentenverteilung anpassen können, um Fahrzeugstabilität und -handling zu verbessern. Die Machbarkeit einer Cloud-Regelung über 4G wird evaluiert, wobei das Potenzial zur Reduktion der Onboard-Rechenleistung gezeigt wird.Human error is the leading cause of road accidents, often resulting from misjudgments of driving conditions or overestimating vehicle capabilities. Advances in vehicle actuation and cloud-connected vehicles offer promising solutions to mitigate such errors by enabling more capable and intelligent vehicle dynamic controllers. This thesis examines how overactuation impacts vehicle handling behaviour, with implications for autonomous vehicle control, and explores how predictive insights into the driving environment, facilitated through vehicle connectivity, enable enhanced proactive vehicle control strategies. An optimisation-based method is proposed to quantify the regions of feasible handling in steady-state conditions, considering actuator sets and limits, thereby defining the achievable vehicle handling behaviour through steering and vehicle sideslip characteristics. Within these regions, Rear-Wheel Steering (RWS), Torque Vectoring (TV) using individual wheel motors, and TV+RWS are evaluated in terms of their impact on power demand, vehicle stability, and friction potential utilisation, and thus maximum normal acceleration. Results indicate that while all three sets can substantially shape characteristics, only TV and TV+RWS provide control over both characteristics independently, with TV+RWS enabling linear steering and vehicle sideslip characteristics, i.e.\ a linear vehicle handling behaviour up to maximum normal acceleration. Exploiting the potential of TV+RWS to enable a largely linear vehicle handling behaviour, a multi-layer architecture is proposed for autonomous path tracking. Here, TV+RWS enforces a linear target vehicle behaviour over a wide range of operation, such that a simplified prediction model becomes feasible for model predictive control, allowing for longer preview horizons and improved computational efficiency by avoiding tyre nonlinearities or overactuation in the prediction model. Analysis and simulations indicate that the architecture not only improves vehicle stability in critical driving conditions but also reduces path tracking errors compared to a baseline vehicle. Towards intelligent vehicle control, the potential of a cloud-based digital twin, which integrates vehicle and environment, is examined to enhance vehicle dynamics control by providing improved vehicle models, enabling proactive control strategies, and supporting personalised vehicle dynamics control in challenging driving conditions. Beyond theoretical discussion, a use case for lateral vehicle stability control is experimentally examined. The experiments demonstrate that utilising information from the driving environment provided by the digital twin, adaptive cruise control, and TV can proactively adjust vehicle velocity and torque distribution to maintain or enhance vehicle stability and handling for drivers. The feasibility of cloud control over public 4G is evaluated, highlighting the potential to reduce onboard computational requirements

    Algorithmic Complexity of Matching and Games

    Full text link
    We study various problems relating to computational social choice (COMSOC) and algorithmic game theory (AGT). Many of the problems we study are NP-hard or beyond. To tackle this complexity, we turn to parameterized complexity and restricted preference domains. Both of these attempt to capture properties of problem instances that make the problems more tractable. In parameterized complexity, we aim to construct algorithms whose exponential components depend only on a specific parameter of the input size. We take a deeper look at a restricted preference domain called d-Manhattan preferences. These preferences arise if the voters and alternatives are located in the d-dimensional real space where the distance is measured using Manhattan distance (aka. l1-norm), and every voter prefers an alternative that is closer to her to one that is further from her. We determine the smallest – in terms of the number of voters and alternatives – preference profiles (collection of voters and alternatives) that are not d-Manhattan, and provide some guarantees on when a preference profile is d-Manhattan. We also study how d-Manhattan preferences relate to other restricted preference domains, such as single-peaked and single-crossing preferences. In the field of matching under preferences, we study the parameterized complexity of refugee resettlement. In this many-to-one matching setting, refugee families must be assigned to places where they can live. The families also have requirements for services, and the places have upper and lower quotas on how much of the services they can provide. We study the parameterized complexity of finding or determining assignments with desirable properties, and obtain a fairly complete picture for multiple natural parameters. Further, we study hedonic games, which can be seen as an extension of matchings. In hedonic games, agents have preferences over coalitions (subsets of agents) containing them. The agents need to be partitioned into disjoint coalitions so that the resulting partition satisfies selected stability constraints. We focus on two compact preference encodings: the model of friends and enemies, and the model of friends, enemies, and neutrals. For both of these models, the preferences can be either friend-oriented or enemy-oriented. We provide a fairly complete parameterized complexity picture for the above-mentioned preference encodings under multiple stability concepts for both friend- and enemy-oriented preferences. Finally, we move to non-cooperative games and study the game implementation problem. In this problem, we are given a normal-form game, some desired outcomes (strategy profiles) of the game, and a budget. The goal is to pay the agents so that they play strategies resulting in the desired outcomes, while we stay within the budget. The problem is already known to be NP-hard; we show that this remains the case in multiple restricted cases, such as when the budget is zero, and there are only two players. We additionally fix a flaw in an earlier exponential-time algorithm for a variant of the problem

    Ionic liquids and their application in molecular diagnostics

    Full text link
    Die molekulare Diagnostik ermöglicht die schnelle und präzise Erkennung, Identifizierung und Charakterisierung mikrobieller und viraler Krankheitserreger. Trotz Fortschritten in den Analysetechniken schränken Herausforderungen wie hohe Kosten, komplexe Ausrüstung, Bedarf von qualifiziertem Personal und Probleme wie komplexe Nukleinsäureextraktion, Inhibition und Kontaminationen die breite Anwendung ein. Ionische Flüssigkeiten (ILs), geschmolzene organische Salze mit einzigartigen Eigenschaften, sind in der Lage Biopolymere zu lösen, Zellen zu lysieren und mit Nukleinsäuren und Proteinen zu interagieren, und bieten vielversprechende Lösungsansätze für oben genannte Herausforderungen. Diese Arbeit untersucht das Potenzial hydrophiler ionischer Flüssigkeiten, um Nukleinsäureextraktions- und Analysemethoden in der molekularen Diagnostik zu verbessern. Eine neuartige DNA- und RNA-Extraktionsmethode für Bakterien, die die ionische Flüssigkeit 1-Ethyl-3-methylimidazoliumacetat und siliziumbeschichtete magnetische Beads verwendet, wurde entwickelt. Die Methode, die IL-basierte Zelllyse bei Raumtemperatur mit Aufreinigung durch magnetische Beads kombiniert, zeigte Effizienzen, die mit kommerziellen Extraktionskits vergleichbar sind, und sparte dabei Aufwand, Zeit und Kosten. Die Methode wurde anschließend für den Nachweis bakterieller Erreger in Urinproben von Patienten mit Harnwegsinfektionen (HWI) angepasst. Der IL-basierte Ansatz ermöglichte eine zuverlässige Detektion mittels qPCR und 16S rRNA-Sequenzierung und bietet eine praktikable Alternative in der UTI-Diagnostik. Die Untersuchung der Lysefähigkeit verschiedener hydrophiler ILs für Pilzsporen und Myzel zeigte außerdem, dass sie schnellere, einfachere und sanftere Alternativen zu gängigen Lysemethoden sind. Zuletzt wiesen Experimente zu den Auswirkungen von ILs auf Nukleinsäuren und molekularen Assays auf ihr großes Potenzial als Enhancer von Nukleinsäureamplifikationsreaktionen hin.Zusammenfassend zeigt diese Arbeit, dass hydrophile ILs ein großes Potenzial haben, das Feld der molekularen Diagnostik für den laborbasierten, on-site oder in-field Einsatz zu verbessern und voranzubringen. Sie bieten die Möglichkeit, Kosten zu senken, die Nukleinsäureextraktion zu vereinfachen und zu beschleunigen, Kontaminationen während der Probenverarbeitung zu verhindern, Inhibitionen entgegenzuwirken und Nukleinsäureanalysereaktionen zu verbessern. Die Integration von ILs in Arbeitsabläufe könnte das Feld der molekularen Diagnostik erheblich verbessern.Molecular diagnostics allow for the rapid and accurate detection, identification, and characterization of microbial and viral pathogens. Despite advancements in analysis techniques, challenges such as high costs, complex equipment, skilled personnel requirements, and issues like complex nucleic acid extraction, inhibition, and contamination limit the widespread application. Ionic liquids (ILs), molten organic salts with unique properties can solubilize biopolymers, lyse cells, and interact with nucleic acids and proteins, offering promising solutions to these challenges. This work explores the potential of hydrophilic ILs to improve nucleic acid extraction and analysis methods in molecular diagnostics. A novel DNA and RNA extraction method for bacteria, using the IL 1-ethyl-3-methylimidazolium acetate and silica-coated magnetic beads was developed. The method, which combines IL-based cell lysis at room temperature with magnetic bead purification, demonstrated efficiencies comparable to commercial kits, while saving effort, time and costs. Consequently, the method was adapted for bacterial pathogen detection in urine samples from patients with urinary tract infections (UTIs). The IL-based approach enabled reliable detection via qPCR and 16S rRNA sequencing, offering a viable alternative in UTI diagnostics. Investigating the lysis potential of various hydrophilic ILs for fungal spores and mycelia also revealed them as faster, simpler and gentler alternatives to common lysis methods. Lastly, research on the effects of ILs on nucleic acids and molecular assays indicated great potential as enhancers of nucleic acid amplification reactions.In summary, this thesis demonstrates that hydrophilic ILs have great potential to enhance and advance the field of molecular diagnostics for lab-based, on-site and in-field applications. They offer the prospect of lowering operational costs, simplifying and accelerating nucleic acid extraction, preventing contamination during sample processing, counteracting inhibition, and enhancing nucleic acid analysis reactions. The integration of ILs into workflows could improve the field of molecular diagnostics significantly

    A Unified Framework for Trend Uncertainty Assessment in Climate Data Records: Demonstration on Global Mean Sea Level

    No full text
    Trends of essential climate variables are often estimated from climate data records to quantify changes in the Earth system. An understanding of the uncertainty in a trend is essential for accurately determining the significance of a trend and attributing its causes. Despite this importance, trend-uncertainty estimates rarely account for all known sources of uncertainty. Common approaches neglect measurement-system instability or neglect the impact of natural variability on trend uncertainty. Such neglect can result in over-confidence in trend estimates. This study addresses trend-uncertainty assessment, particularly the need to account for the combined effects of measurement instability and natural variability on the trend uncertainty. The study presents a novel, unified framework for trend estimation that combines available measurement uncertainty information with empirical modelling of natural climate variability to achieve a more accurate uncertainty estimate. The framework is demonstrated for a time series of global mean sea level observations, obtaining more realistic trend-uncertainty values. The framework is applicable to most other climate data records. Adopting this approach will enhance confidence in climate change analysis through more accurate trend-uncertainty assessment in climate studies

    29,302

    full texts

    256,869

    metadata records
    Updated in last 30 days.
    reposiTUm is based in Austria
    Access Repository Dashboard
    Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇