openHSU (HSU Hamburg)
Not a member yet
7291 research outputs found
Sort by
Artificial Intelligence in Security and Defense
No full textThis book is licensed under the terms of the Creative Commons Attribution 4.0 International License, which means you are free to use, share, adapt, distribute, and reproduce it in any medium or format, as long as you give appropriate credit to the original author(s). You can find more information at http://creativecommons.org/licenses/by/4.0/Artificial Intelligence is increasingly recognized as a dual-use technology, where the same solutions are applied to both civil security and defense. However, its integration into these domains extends beyond the mere adoption of existing AI methodologies. Security and defense introduce distinct and complex requirements, particularly in areas such as ethics and cybersecurity, which necessitate tailored AI solutions. These requirements, alongside potential AI-driven approaches for security and defense, were central themes at the 1st workshop on AI4SD - Artificial Intelligence in Security and Defense – at the European Conference in Artificial Intelligence in Bologna held on October 26th, 2025.Vo
A vision on the potential of AI in human expert input and labeling in military lessons learned
No full textEffective Knowledge Management (KM) is critical for ensuring that the right information reaches the right personnel at the right time. This process begins with the capture and storage of high quality information to avoid the common trap of ‘Garbage in = Garbage out.’ This paper examines the North Atlantic Treaty Organization’s (NATO) Lessons Learned Process, as a KM system that includes a change management process, emphasizing its role in enhancing interoperability and Command and Control within NATO military headquarters. The quality of the input and labeling of the observations and lessons captured, stored, and shared during the process contributes significantly to the likelihood that the Lessons Learned Process will result in tangible and lasting change. Current approaches to human-expert input and labeling do not ensure the necessary quality. This paper explores the potential of Artificial Intelligence (AI) techniques to augment or replace human involvement in these tasks. We present a forward-looking perspective on the integration of advanced AI and Machine Learning (ML) technologies to elevate the quality and effectiveness of the input and labeling of information during the Lessons Learned Process.Vo
Wargaming in the age of AI
No full textWargames are regarded as a suitable instrument through which practitioners may anticipate the complexity and unpredictability of future lived experience by means of ludic procedures. Over the past decade, wargaming has not only experienced a renaissance within military contexts but has also been rearticulated as a prioritized tool for learning and training. In this regard, the integration of technologies from the field of Artificial Intelligence (AI) —particularly Large Language Models (LLMs)—is expected to assist in managing the emergent contingencies of modern warfare. Building on the promises attached both to military wargaming and to the implementation of LLMs within such practices, this contribution focuses, on the one hand, on the epistemological foundations of wargaming, with and without computational augmentation. On the other hand, by outlining the ethical risks implied by these automation strategies, it raises the question of what consequences the integration of technical agency entails for the pedagogical aspirations of wargaming and for ethically informed decision-making processes. In this context, the article further interrogates how the epistemological status of wargaming itself is being transformed through the turn to automation.Vo
Adaptive uncertainty quantification for maritime classification under cloud cover in satellite imagery
No full textClassifying vessels in satellite imagery is important for naval intelligence, search-and-rescue, and the monitoring of illegal activities. Performance, however, degrades under adverse meteorological conditions (e.g., clouds), which obscure discriminative features and increase ambiguity of class membership. Conventional deep learning methodologies fail to directly address this ambiguity, sometimes generating overly confident predictions for a specific erroneous class. To address this gap, we present WAVES (Weather-Aware Visual Estimation with Sets), a conformal prediction framework that makes uncertainty quality-aware. WAVES learns a temperature map that adjusts classifier logits as a function of predicted cloud coverage. A single global conformal threshold is then computed using the standard order statistic, preserving split-conformal marginal coverage while producing prediction sets that adapt across quality conditions.
We evaluate WAVES using a publicly accessible dataset comprising various ship classes, which we augment with realistic synthetic clouds. Across three relevant classification models from the literature and compared against standard split–conformal prediction, WAVES typically maintains or improves coverage while reducing prediction–set size; across all three models, it reduces average set size by 6–7% on average (up to 15% in the best setting) while keeping coverage within ±0.1 percentage points of the global split–conformal prediction baseline. These results indicate that quality-aware calibration combined with a global conformal threshold provides reliable, compact prediction sets for maritime surveillance under variable cloud cover, especially in operational scenarios where classification outcomes directly influence critical human or autonomous decisions.Vo
A declarative approach to strategic deconfliction in urban air mobility
No full textThe increasing demand for Urban Air Mobility (UAM) has introduced a new layer of complexity in managing airspace, particularly in densely populated metropolitan areas. As the number of air vehicles, including drones, air taxis, and helicopters, continues to grow, the risk of mid-air collisions and conflicts with other air traffic and obstacles increases. Strategic deconfliction is critical to ensuring safe and efficient UAM operations. This paper proposes an Answer Set Programming (ASP) solution to real-world strategic deconfliction problem in UAM with respect to time synchronization and optimization of the flight route so that all flights are in deconfliction. The ASP solution is compared with a Constraint Programming (CP) approach, investigating the efficiency of both approaches through scalability tests and comparing the respective time and memory requirements. The results show that ASP generally offers faster execution and better scalability for small to medium-sized problems, while CP exhibits a more consistent memory usage but struggles significantly with the execution time as the problem complexity increases.Vo
Fab City Hamburg Playbook
No full textErkenntnisse aus dem ReallaborDeutsche Online VersionDas Fab City Hamburg Playbook fasst Erkenntnisse aus dem transdisziplinären Projekt „Fab City: Dezentrale digitale Produktion für die urbane Wertschöpfung“ zusammen, dass unter der Leitung des New Production Institute an der Helmut-Schmidt-Universität durchgeführt wird. Im Rahmen des Projektes wurde seit 2020 ein umfangreiches Reallabor in Hamburg aufgebaut, dass Forschende unterschiedlicher Disziplinen, Institutionen aus dem öffentlichen und privaten Sektor und Nutzerinnen zusammenbringt, um gemeinsam an Lösungen für die produktive Stadt der Zukunft zu arbeiten. Im Fokus steht die Transformation städtischer Produktions- und Konsumptionsmuster hin zu einer urbanen Kreislaufwirtschaft, und die Frage was und wie wir lokal produzieren können, um den Weg zu nachhaltigen und resilienten Produktionssystemen zu ebnen.
Das Playbook richtet sich an ein breites Publikum aus Praktikern, Forschenden und interessierten Hamburger Bürgerinnen. Es erzählt und dokumentiert aus dem Projekt erwachsene Stadtgeschichten von erfolgreichen Prototypen und kreativen Köpfen, die als Wegbereitende der Transformation den Wandel gestalten. Es hält konkret und praxisbezogen unsere Erfahrungen im Reallabor Hamburg fest und soll als Informations- und Inspirationsquelle dienen.Vo
Reactive temperament and self-regulation in preschool
No full textThis is an Open Access article distributed under the terms of the creative commons Attribution license (http://creativecommons.org/licenses/by/4.0/)HSU-O
Assessing robustness in data-driven modeling of cyber-physical systems
No full textRobustness is a key factor in the design and analysis of Cyber-Physical Systems (CPS), ensuring that systems function correctly even under perturbations. This paper investigates robustness within the data-driven modeling process, focusing on three core aspects: system robustness, model robustness, and learner robustness. We survey existing notions of robustness and propose unified formal definitions for each aspect, analyzing their interdependencies and their contributions to overall CPS performance. Additionally, we introduce a method for assessing the robustness of models generated by data-driven learning that is independent of both the model’s internal representation and the learning paradigm used. Our approach leverages input perturbations combined with probabilistic analysis to evaluate how well a learned model handles input variations, particularly when formal guarantees are challenging to obtain. To demonstrate the practical application of our method, we conduct a case study on a temperature control system, using decision trees to model system behavior. By perturbing test data and analyzing the resulting model outputs, we identify non-robust regions near decision boundaries, thereby revealing potential vulnerabilities. The proposed framework offers valuable insights for enhancing system design and lays the groundwork for future research into robust machine learning models for CPS.Vo
High-power, ultrafast source for XUV frequency comb spectroscopy
No full textThis work is devoted to creation of a unique laser source for high-precision spectroscopy in the extreme ultraviolet (XUV) spectral region, where no continuous wave lasers exist. The optical region contains a vast of intriguing atomic and electron transitions that still need to be investigated, such as an isomeric nuclear transition in 229Th and the 1S-2S transition in trapped He+ ions. Namely, the work focuses on developing a high average and peak power laser system at a multi-megahertz repetition rate, which should drive high harmonics in a noble gas to transfer the optical frequency comb into the short wavelength region. Since the conversion efficiencies in high harmonic generation processes typically do not exceed ~10^-6, a driver should possess high average power and energy in combination with low noise. While high pulse peak power could improve the conversion efficiency, the high average power would ensure reasonable power in the generated XUV comb lines for spectroscopy applications. Thin-disk Kerr-lens mode-locked oscillators were chosen as robust and powerful femtosecond sources possessing high peak and average powers with the proven potential for scalability. In combination with a new approach of nonlinear spectral broadening and pulse compression in Herriott-type multipass cells, the thin-disk oscillators can provide the necessary peak power (>1 GW) in combination with short pulses to efficiently drive high harmonics in gas.
The work is divided into two packages. Firstly, a powerful Kerr-lens mode-locked thin-disk oscillator was developed. Thanks to the scalability of this type of oscillators, it was possible to reach the unprecedented output parameters in a few iterative steps. The final version of the oscillator set a world record on output peak power, namely 110 MW, corresponding to 115 fs-long pulses and 202 W average power at a 14 MHz repetition rate. Additionally, it was demonstrated that the scaling of this type of oscillator can be well predicted and realized beyond 100 MW peak powers.
The second part of the thesis focused on boosting the pulse peak power of the developed oscillator for subsequent high harmonic generation in gas jets. To achieve this, the oscillator output pulses underwent nonlinear spectral broadening and pulse compression in Herriott-type multipass cells. As a proof of principle, pulse compression was initially demonstrated with a commercial Pharos laser at reduced average power but equivalent to the oscillator peak power. Finally, the approach was transferred to the high peak- and average power oscillator. The 120 fs pulses were compressed in two cascaded multipass cells by a factor of 15 down to 8.0 fs, corresponding to 148 W average power, 0.9 GW pulse peak power with 82% overall throughput. Additionally, a sub-two-cycle operation with the compressed 6.2 fs long pulses was demonstrated. The multipass cells relying on the all-dielectrically coated mirrors and gas as a nonlinear medium proved to be highly suitable as well as scalable for spectral broadening and compression of high average- and peak power Yb-based lasers.
Moreover, several proof-of-principle applications of the developed systems were demonstrated. Among them were mid-infrared generation spanning the range of 2 – 20 µm, multiphoton imaging of the biological tissues, and the first experiments on high harmonic generation.Diese Arbeit widmet sich der Entwicklung einer einzigartigen Laserquelle für die hochpräzise Spektroskopie im extrem ultravioletten (XUV) Spektralbereich, für den bisher keine Dauerstrichlaser existieren. Dieser Wellenlängenbereich enthält eine Vielzahl bedeutender optisch anregbarer Übergänge in der elektronischen Hülle mit noch ungenau bestimmten Anregungsenergien, wie bspw. im Fall des 1S-2S-Überganges in gefangenen He+ -Ionen. In besonderen Fällen ist es auch möglich, niedrigenergetische Übergänge im Atomkern optisch anzuregen, wie z. B. im Falle des isomeren Kernüberganges in 229Th. Die Arbeit konzentriert sich auf die Entwicklung eines Lasersystems mit hoher Durchschnitts- und Spitzenleistung bei einer Wiederholrate von mehreren Megahertz. Hiermit sollen höheren Harmonische in einem Edelgas erzeugt werden, um einen optischen Frequenzkamm in dem XUV-Bereich zu realisieren. Da die Umwandlungseffizienzen bei der Erzeugung hoher Harmonischer in der Regel ~10^-6 nicht überschreiten, sollte das Lasersystem eine hohe mittlere Leistung und Energie in Kombination mit geringem Rauschen aufweisen. Während eine hohe Pulsspitzenleistung die Umwandlungseffizienz verbessern könnte, würden hohe Durchschnittsleistungen eine ausreichende Leistung der erzeugten XUV-Kammlinien für Spektroskopieanwendungen gewährleisten. Als robuste und leistungsstarke Femtosekunden-Strahlquellen mit hoher Spitzen- und Durchschnittsleistung und nachgewiesener Skalierbarkeit wurden Kerr-Linsen-modengekoppelte Dünnscheibenoszillatoren ausgewählt. In Kombination mit einem neuen Ansatz der nichtlinearen spektralen Verbreiterung und Pulskompression in Herriott-Multi-Pass-Zellen können die Dünnscheibenoszillatoren die erforderliche Spitzenleistung (>1 GW) in Kombination mit kurzen Pulsen liefern, um hohe Harmonische in Gas effizient zu erzeugen.
Die Arbeit ist in zwei Teile unterteilt. Erstens wurde ein leistungsfähiger Kerr-Linsen-modengekoppelter Dünnscheibenoszillator entwickelt. Dank der Skalierbarkeit dieses Oszillatortyps war es möglich, bisher unerreichte Ausgangsparameter in wenigen Iterationsschritten zu erreichen. Die endgültige Version des Oszillators stellt einen Weltrekord bei der Spitzenleistung von 110 MW dar, mit 115 fs langen Laserpulsen und einer Durchschnittsleistung von 202 W bei einer Wiederholrate von 14 MHz. Außerdem wurde gezeigt, dass die Skalierung der Spitzenleistung mit diesem Oszillatortyp bisher erwartungsgemäß realisierbar gewesen ist und daher anzunehmend ist, dass eine Skalierung jenseits von 100 MW Spitzenleistung realistisch erscheint.
Der zweite Teil der Arbeit konzentrierte sich auf die Erhöhung der Pulsspitzenleistung des entwickelten Oszillators für die anschließende Erzeugung hoher Harmonischer in Gasstrahlen. Um dies zu erreichen, wurden die Laserpulse des Oszillators einer nichtlinearen spektralen Verbreiterung und zeitlichen Kompression in Herriott-Multi-Pass-Zellen unterzogen. Zum Beweis des Prinzips wurde die Pulskompression zunächst für einen kommerziellen Pharos-Laser bei reduzierter mittlerer Leistung, jedoch identischer Spitzenleistung, demonstriert. Schließlich wurde der Ansatz auf den Oszillator mit hoher Spitzen- und Durchschnittsleistung übertragen. Die 120 fs-Pulse des Laseroszillators wurden in zwei kaskadierten Multi-Pass-Zellen um den Faktor 15 auf 8,0 fs komprimiert, was einer mittleren Leistung von 148 W und einer Pulsspitzenleistung von 0,9 GW bei einem Gesamtdurchsatz von 82% entspricht. Außerdem wurde ein Betrieb mit auf 6,2 fs komprimierten Laserpulsen demonstriert, welche somit weniger als zwei Feldoszillationen einschließen. Die Multi-Pass-Zellen, die ausschließlich dielektrisch beschichtete Spiegel und Gas als nichtlineares Medium enthalten, erwiesen sich als äußerst geeignet für die spektrale Verbreiterung und Kompression von Yb-basierten Lasern mit hoher Durchschnitts- und Spitzenleistung. Diese Multi-Pass-Zellen lassen sich noch zu höheren Pulsspitzenleistungen und Durchschnittsleistungen skalieren.
Darüber hinaus wurden zu Demonstrationszwecken mehrere Anwendungen der entwickelten Systeme aufgezeigt. Dazu gehören die Erzeugung von Licht im mittleren Infrarotbereich von 2 – 20 µm, die Multiphotonen-Bildgebung von biologischem Gewebe, sowie erste Experimente zur Erzeugung von höheren Harmonischen im XUV-Bereich.Vo
Corrigendum to “FSI simulations of wind gusts impacting an air-inflated flexible membrane at Re = 100,000” [J. Fluids Struct. 109 (2022) 103462]
No full textThe authors regret the presence of a typographical error in Eq. (1) of De Nayer et al. (2022). As correctly introduced in De Nayer and Breuer (2020), the source term used to inject the wind gust should be analogous to a mass flow rate multiplied by the total flow velocity [Formula presented] and not only the gust velocity component [Formula presented]. Therefore, the source term written in the Cartesian basis [Formula presented] should read: [Formula presented] The authors would like to apologize for any inconvenience caused. CRediT authorship contribution statement G. De Nayer: Writing – original draft. M. Breuer: Resources, Supervision. K. Boulbrachene: Writing – original draft.Vo