Recent Submissions
Bildverarbeitungs- und Planungsmethoden für intelligente Reinigungsroboter in Büroumgebungen
(Stuttgart : Fraunhofer-Institut für Produktionstechnik und Automatisierung IPA, 2026) Bormann, Richard Klaus Eduard; Verl, Alexander (Univ.-Prof. Dr.-Ing. Dr. h. c.)
Aufgrund von Fachkräftemangel, Mindestlohnregelungen, Kostendruck und technologischer Weiterentwicklung steigt in der professionellen Gebäudereinigungsbranche der Automatisierungsdruck zunehmend. Mittlerweile sind gewerbliche Reinigungsroboter am Markt verfügbar, welche in großflächigen Umgebungen wie Supermärkten, Turnhallen oder Lagern gemäß vordefinierter Bewegungsmuster die Flächenreinigung durch Nasswischen, Staubsaugen oder Kehren autonom übernehmen. Für Büroumgebungen sind wegen der deutlich anspruchsvolleren Randbedingungen und komplexeren Reinigungsprozesse dagegen noch keine kommerziellen Roboterlösungen verfügbar. Analysen haben ergeben, dass für den Einsatz in der täglichen Unterhaltsreinigung allein in Deutschland ein Bedarf an über 10.000 Robotereinheiten besteht. Diese sollen vor allem die Bodenreinigung und Papierkorbentleerung autonom übernehmen - zwei Tätigkeiten, welche 70% der täglich anfallenden Aufwände ausmachen - während sich das Reinigungspersonal auf anspruchsvollere Aufgaben wie die Sanitär- und Oberflächenreinigung konzentrieren kann.
Im Rahmen der vorliegenden Arbeit soll eruiert werden, ob durch entsprechend intelligente, softwaretechnische Funktionsmodule zur Navigationsplanung und Bildverarbeitung professionelle autonome Reinigungsroboter soweit zur Autonomie und robusten Aufgabenbewältigung befähigt werden können, dass unter den üblichen Randbedingungen von Dienstleistungen in der Reinigungsbranche ein zuverlässiger und wirtschaftlicher Einsatz von Servicerobotern für komplexere Bodenreinigungstätigkeiten und einfache Handhabungsvorgänge zur Abfallentsorgung ermöglicht wird.
Dazu erforscht, entwickelt und analysiert diese Arbeit intelligente Algorithmen für die automatische Einteilung von Büroumgebungen in sinnvolle Arbeitsbereiche sowie Ansätze für die autonome Bewegungsplanung zur flächenabdeckenden Bodenreinigung und zur Gesamtplanung optimierter Reinigungsabfolgen im Gebäude. Diese werden komplementiert durch Algorithmen für die visuelle Erkennung von Verschmutzungen und Büroobjekten für die zielgerichtete Reinigung auf den zu reinigenden Bodenflächen sowie Methoden zur Detektion und Handhabung von Papierkörben für die Entleerung.
Für die automatische Grundrisseinteilung und die flächenabdeckende Bewegungsplanung werden neue, ressourcensparsame Methoden für den Einsatz auf mobilen Robotern entwickelt und umfangreich mit einer repräsentativen Auswahl verschiedenster klassischer Ansätze auf einer neu entwickelten Benchmark-Datenbasis verglichen. Nach Wissensstand des Autors ist diese Arbeit eine der wenigen, welche überhaupt eine strukturierte, komparative Analyse verschiedener Verfahren durchführt und hier Orientierung bietet. Die Arbeitsabfolgeplanung über das gesamte Gebäude erfolgt mit verschiedenen heuristischen und exakten Optimierungsmethoden und wird im Zusammenspiel mit der Grundrisseinteilung empirisch bezüglich aller Parameter ausgewertet. Für die visuelle Schmutzerkennung werden ein modellfreier und ein angelernter Detektionsalgorithmus sowie ein Datengenerator für die Erzeugung beliebig vieler Datenbeispiele vorgestellt, zwei Datensätze aufgenommen und die Verfahren darauf eingehend untersucht. Ein ähnlicher Objektdetektor wird für die Erkennung von Papierkörben für die Entleerung entwickelt und mitsamt einem entsprechenden Datensatz ausgewertet. Die entwickelten Methoden wurden im Rahmen zweier Forschungsprojekte auf drei verschiedenen Robotersystemen in Labor- und Kundenumgebungen der Firmen Kenter und Dussmann erfolgreich evaluiert und sind somit geeignet, in zukünftigen gewerblichen Reinigungsroboterprodukten zum Einsatz zu kommen. Zur Verwertung und Weiterentwicklung sind die meisten Implementierungen und Datensätze öffentlich verfügbar und werden bereits aktiv von einer breiten Nutzerschaft verwendet.
Targeted exploration and robust dual control for linear dynamical systems
(2026) Venkatasubramanian, Janani; Allgöwer, Frank (Prof. Dr.-Ing. Dr. h.c.)
Vehicle communication in cellular automata : simulation of bidirectional traffic in motorcycle platoons
(2026) Boike, Nils
This thesis investigates the use of cellular automata to model communication-enhanced motorcycle platoons in bidirectional traffic scenarios. The central research question addresses how decentralized communication between motorcycles, and with opposing traffic, influences safety and traffic behavior within a discrete traffic simulation model. While previous cellular automata traffic models have successfully captured traffic dynamics, they have largely focused on car-based traffic and local perception, failing to address cooperative communication in motorcycle platoons. To address this gap, this work extends a cellular automata traffic model with communication-based anticipation mechanisms tailored to the characteristics of motorcycle platoons. The approach is evaluated through a simulation-based methodology, including the implementation of a custom cellular automaton model, multiple communication scenarios, quantitative safety and traffic flow metrics, and qualitative time-space analysis. The results show that intra-platoon communication improves safety and stability, while broader communication scopes exhibit mixed effects.
Targeting glioblastoma mitochondrial metabolism with S-Gboxin induces cytotoxicity under conditions of the tumor microenvironment
(2026) Weinem, Jan-Béla; Urban, Hans; Sauer, Benedikt; Buhlmann, Tanja; Hau, Ann-Christin; Liebner, Stefan; Rusch, Tillmann; Namgaladze, Dmitry; Harwart, Leander F.; Schröder, Jan-Hendrik; de Souza, Maeve; Steinbach, Joachim P.; Legewie, Stefan; Luger, Anna-Luisa; Ronellenfitsch, Michael W.
Glioblastoma (GB) is the most common primary malignant brain tumor in adults. Gboxin, a novel compound that targets oxidative phosphorylation via complex V inhibition, has shown promise in preclinical models of GB. We examined the efficacy of the pharmacokinetically optimized S-Gboxin under conditions replicating the GB microenvironment, including nutrient deprivation and hypoxia. We assessed cytotoxicity and growth-inhibitory effects of S-Gboxin in human GB cell lines, primary GB cultures, as well as immortalized and primary human astrocytes under different nutrient and oxygen deprivation scenarios. Oxygen consumption, cell migration, activation of the integrated stress response (ISR) as well as the relevance of the AMP-activated protein kinase (AMPK) were evaluated as variables under S-Gboxin treatment. S-Gboxin demonstrated cytotoxicity at low micromolar concentrations, with cell death enhanced under nutrient deprivation and hypoxia. S-Gboxin reduced cellular oxygen consumption and uncoupled mitochondria. Cytotoxicity was increased when mitochondrial fuels were the primary energy source. Additionally, S-Gboxin treatment resulted in elevated lactate production and glucose consumption. While the ISR marker ATF4 was induced by S-Gboxin in a dose-dependent manner, ISR inhibition with ISRIB did not affect its cytotoxicity. Conversely, S-Gboxin treatment combined with AMPK inhibition resulted in enhanced tumor cell death. Collectively, these findings demonstrate that S-Gboxin selectively targets cancer-specific metabolic vulnerabilities in GB cells. The synergistic action with AMPK inhibition suggests that this pathway contributes to maintain energy homeostasis in the presence of the drug. Therefore, S-Gboxin is a promising compound for GB therapy, especially in a combinatory approach with AMPK inhibition or other metabolic targeted therapies.
Microaerobic dark fermentation by purple bacteria as an emerging perspective for biohydrogen production : a review
(2026) Krake, Simon; Tovar, Günter E. M.; Zibek, Susanne
While in recent years, a lot of research regarding microbial dark fermentation for biohydrogen production has been carried out focusing on improving well-established processes, this review aims to draw attention to the discovery of a new method of process control, i.e., microaerobic dark fermentation in purple non-sulfur bacteria. This new subsegment of research tries to rethink efficiencies of biohydrogen production and substrate conversion based on an increasingly comprehensive understanding of metabolic pathways, particularly in the area of interplay between aerobic and anaerobic metabolism and the control of this state. By combining hydrogen production through dark fermentation and partially activated photofermentation by a partially active nitrogenase (~ 12%), microaerobic fermentation has the potential to increase hydrogen efficiency. At the same time, it allows to maintain advantages of facultative aerobes, such as high-cell densities and growth rates. Studies show that very high yields of up to 8-12 mol H2 mol-1 substrate should theoretically be possible with this method, but the yields currently realized are still in the range already achieved with dark fermentation (0.2-1.6 mol H2 mol-1 substrate). This review provides an overview of the relevant bacteria, metabolism, cultivation, and challenges to further increase hydrogen productivity using microaerobic dark fermentation by purple bacteria. In addition, the model microorganism Rhodospirillum rubrum is considered in more detail to gain a deeper understanding of the processes at the metabolic level under microaerobic conditions. The review concludes with a detailed proposal of the challenges and opportunities to create a new, exciting perspective for biotechnological hydrogen production by microaerobic dark fermentation.
A lazy and modular approach to int-blasting
(2026) Barth, Max; Heizmann, Matthias; Hoenicke, Jochen
Bit-vector operations are ubiquitous in programming languages and formal verification, but their complex semantics pose challenges for SMT solvers. Although bit-blasting-translating bit-vectors to Boolean variables-is widely used, it struggles with arithmetic bit-vector operations on large bit-widths (e.g., 64-bit or 256-bit variables) due to exponential blowup. Int-blasting, which maps bit-vectors to integer arithmetic, offers a scalable alternative for arithmetic bit-vector operations, but introduces many modulo operations of which some are redundant. This article presents a modular three-step translation from bit-vector formulas to integer formulas, designed to keep the amount of modulo operations low, while preserving correctness. In the first step, we translate bit-vector operations to integer operations. Thereby, we introduce the two functions and as explicit operators in the SMT-LIB theory of bit-vectors. Each integer operation is wrapped by and . Hence, the sort of all bit-vector terms is preserved. Therefore, the first translation step is an equivalence transformation. In the second step, we simplify the formula by replacing the composition with a modulo operation. These modulo operations are added lazily, i.e., if the modulo does not change the result of the operation, it is omitted. In our experiments this reduced the average amount of modulo operations by 51%. In the third step, we introduce lemmas to precisely capture the meaning of and . We prove that these lemmas suffice to solve bit-vector formulas. Furthermore, we illustrate that these lemmas are also sufficient for bit-vector formulas with quantifiers, arrays and uninterpreted functions. We implement our translation in SMTInterpol and evaluate it on 19570 SMT-LIB benchmarks. Results show that our lazy int-blasting solves 15% more tasks than an eager int-blasting, with 35% faster average runtime and 12% lower memory usage.
Local thermal non-equilibrium models in porous media : a comparative study of conduction effects
(2026) Kostelecky, Anna Mareike; Stefansson, Ivar; Bringedal, Carina; Ghosh, Tufan; Dahle, Helge K.; Helmig, Rainer
Instantaneous heat transfer between different phases, known as local thermal equilibrium (LTE), is commonly assumed for modeling heat transfer in porous media. This assumption may not hold in certain technical and environmental applications, particularly with large temperature gradients, large differences in thermal properties, or high velocities. Local thermal non-equilibrium (LTNE) models aim to describe heat transfer processes when the LTE assumption may fail. We compare three continuum-scale models from the pore to the representative elementary volume (REV) scale. Specifically, dual-network and REV-scale models are evaluated against a pore-resolved model used as a reference in the absence of experimental results. Different effective models are used to obtain upscaled properties on the REV scale and to compare resulting temperature profiles. The systems investigated are fully saturated, consisting of one fluid and one solid phase. This study focuses on purely conductive systems without significant differences in thermal properties. Results show that LTE holds for low interfacial resistances. However, for large interfacial resistances, solid and fluid temperatures differ. The REV-scale model with effective parameters obtained by homogenization leads to similar results as the pore-resolved model, whereas the dual-network model deviates more due to fixed spatial resolution. Among the evaluated REV-scale formulations, only the homogenization-based approach captures the LTNE behavior, as it incorporates the interfacial heat transfer coefficient. Our results provide a basis for conduction-dominated heat transfer in saturated porous media and for further systematic comparisons that incorporate convection relevant to a broader range of applications.
Synchronization primitives as the distributed ready queue
(2026) Floor, Ramon
Modern software increasingly relies on cooperative multitasking. With cooperative multitasking, scheduling and synchronization is handled entirely in user space. This avoids heavy system calls and thread-level context switches, which benefits performance. In state-of-the-art user space scheduled systems, scheduling and synchronization is tackled separately. Scheduling is handled by a work-stealing thread pool and synchronization is done with task-aware synchronization primitives. This design, however, can introduce additional overhead under high contention.
In this thesis, we want to investigate a new approach to user space scheduling that combines scheduling and synchronization into one system. Instead of maintaining a global ready queue at the thread pool, we distribute ready queues among the mutexes, such that each mutex has a blocked and a ready queue. Under high contention, critical sections often show the pattern that when a task leaves the critical section, another task is already attempting to enter it and is consequently blocked. With this design we can leverage this observation and hand control from the entering task that is blocked to the one that left the critical section. We call this approach Relay Scheduling. We compare our new approach to existing solutions using several microbenchmarks as well as an application benchmark. We will show that Relay Scheduling performs particularly well for low thread counts.
A data plane interface for resource-constrained microcontrollers in time-sensitive networking
(2026) Kupka, Bastian
Time-Sensitive Networking (TSN) enables deterministic Ethernet communication through coordinated transmission control. While there are dedicated TSN switches and network interface cards, they are rarely available on resource-constrained microcontrollers. In particular, transmit-time-based scheduling approaches such as the Earliest TxTime First (ETF) queuing discipline are typically not supported on embedded platforms.
This thesis investigates the feasibility of implementing ETF-like transmit-time scheduling on a microcontroller using Zephyr RTOS. The implementation targets the NXP i.MX RT1062, which provides a precise PTP hardware clock and compare interrupt mechanism but lacks dedicated hardware support for traffic shaping. The proposed solution integrates transmit-time scheduling into the Zephyr networking stack by combining hardware-triggered interrupts with software-based buffer management inside the network driver.
Scheduled frames are prepared in advance and transmitted using a PTP compare event to trigger the transmission routine close to the target time, followed by a short busy-wait phase to improve precision. Best-effort traffic is handled through a driver-level guard band to avoid interference with scheduled transmissions.
Experimental evaluation shows that, for periodic traffic with fixed inter-packet gaps, the implementation achieves a bounded transmission window of approximately 3 µs. For variable inter-packet gaps, a larger timing spread is observed. The results demonstrate that ETF-like transmit-time scheduling can be realized on a low-cost microcontroller for certain traffic patterns by leveraging existing PTP hardware features.
Towards supersolids of laser-cooled dipolar molecules
(2026) Groß, Phillip; Pfau, Tilman (Prof. Dr.)
Quantum gases of diatomic molecules are many body systems that offer unprecedented control over their intrinsic long and short range interactions. This is due to the unique nature of the interactions between molecules, which can be much stronger than those in magnetic atoms and which can be tuned independently through external electromagnetic fields. As a result, diatomic molecules enable both the exploration of novel supersolid states and the creation of dipole dominated and self assembled quantum solids.
The molecular properties and internal structure of calcium monofluoride (CaF) make it a promising candidate molecule for the realization of several kinds of strongly correlated quantum systems, such as strongly dipolar Bose Einstein condensates, quantum crystals and arrays of molecules in optical tweezers. This dissertation summarizes the first steps towards the realization of such systems using directly laser cooled CaF molecules. The two major experimental steps which are summarized in this thesis are cryogenic buffer gas cooling and transversal laser cooling of CaF molecules.
For the first achievement, cryogenic buffer gas cooling, it is important to observe that CaF molecules are radicals and therefore do not exist in a stable gas phase. As a result, the production of a beam of molecules requires a cryogenic buffer gas cell setup, in which the molecules are produced via laser ablation of a solid precursor target. Two different kinds of ablation target are used to produce a molecular beam. The first is a pellet of a mixture of CaF2/BaF2. The second is a calcium metal target in addition to sulfurhexafluoride gas, which acts as a fluorine donor. The metal target is found to be the much more stable molecule production method, and the buffer gas cooling setup produces a bright and slow beam of molecules with a flux of around 1011 - 1013sr-1 molecules in the ground state of the laser cooling cycle.
The second achievement is a significant improvement of the techniques for laser cooling of the resulting molecular beam. Notably, trapped cold gases of directly laser cooled molecules have not reached the quantum degenerate regime yet, but this milestone is within reach. In our experiment we aim to improve the loading of the magneto optical trap from the laser slowed beam, as this has been the most inefficient step in previous experiments with CaF. Our approach is to collimate the molecular beam through transversal laser cooling on the 𝑋 - 𝐴 transition. Due to the angular momentum structure of the type-II cooling transition, Doppler cooling forces in molecules are typically accompanied by sub Doppler Sisyphus heating forces and vice versa. This has made the collimation of molecular beams difficult in previous experiments, because efficient Doppler cooling over a large velocity range is perturbed by the appearance of sub Doppler heating forces at low transversal velocities.
Here, we present a new way to address the hyperfine structure in the molecules using carefully chosen laser sidebands, which overcomes this limitation. The sideband spectrum of the cooling laser consists of one strong sideband with ≈ 80% of the laser power addressing a single hyperfine state and several weaker sidebands which repump population from the other hyperfine states. Depending on the detuning of the cooling laser, the strong sideband generates sub Doppler cooling forces on the 𝑋2Σ+(𝐹 = 2) - 𝐴2Π1/2 (𝐹 = 1) or the 𝑋2Σ+(𝐹 = 1+) - 𝐴2Π1/2 (𝐹 = 0) transition. The measured transversal density profiles of the molecular beam suggest that, for certain detunings of the cooling laser, both mechanisms lead to forces that counteract the molecular motion, leading to the simultaneous appearance of Doppler and sub Doppler cooling. This result is contrary to the conventional tendency of type II systems, where sub Doppler cooling forces are usually opposed by Doppler heating forces. The observed behavior is supported by numerical simulations of the laser cooling forces and molecular trajectories based on the optical Bloch equations. Overall, the results highlight the importance of a well tailored cooling laser sideband structure for molecular laser cooling to control the interplay between both cooling mechanisms.
Through the collimation of the molecular motion towards the capture region of a magneto-optical trap, the transversal cooling of the molecular beam with simultaneous Doppler and sub-Doppler cooling forces should enable a more efficient trap loading. The experimental results presented here therefore form the basis for future experiments on laser slowing, magneto optical trapping and evaporative cooling towards Bose Einstein condensation or the loading of large optical tweezer arrays of cold CaF molecules.