335598 research outputs found
Sort by
Industry-relevant benchmark dataset for the evaluation of 3D reconstruction methods and evaluation of different NeRF methods
No full textThis study assesses the potential of different Neural Radiance Field (NeRF) and 3D Gaussian Splatting (3DGS) methods in terms of the rendered images and extracted point clouds against classical Multi View Stereo (MVS) with a focus on industrial objects. Firstly, a benchmark dataset is created and captured with an industrial robot arm and camera setup. The camera poses are derived in two alternative versions, which are compared with each other: with Structure from Motion (SfM) and with the robot, by utilizing the kinematics of the robot and the hand-eye pose. The robot-derived poses are more robust and independent of the captured scene. Moreover, they are metric and can be determined much faster than the SfM poses. Secondly, current NeRF and 3DGS methodsvare evaluated on the benchmark data set, each with both pose versions. The results show that with both pose versions, similarly accurate images from novel views can be rendered. However, the images rendered with SfM poses are less accurate for complex objects. The extracted point clouds of trained NeRF and 3DGS models are usually less accurate than standard MVS point clouds, except for some transparent object
Architecture in the Cradle: Early Warning of Architectural Decay with ArchGuard
No full textArchitectural decay can manifest as the evolution of architectural smells, degrading integrity, and increasing maintenance costs. Existing techniques capture smells post hoc or predict on component level, acting too late or on too coarse a granularity. We investigate if the risk of introducing architectural smells can already be predicted when issues are opened. Thus, we propose an issue-level prediction approach that utilizes the semantic representations of Large Language Models (LLMs). To enable training and evaluation, we construct a dataset from three GitLab-hosted projects by linking issues to smells via smell-inducing changes. On this dataset, we train classifiers to identify high-risk issues and conduct an empirical study comparing seven different representations and nine classifiers. Our best-performing classifier (SVM with OpenAI embeddings) achieves F1-scores of up to 0.506, with a recall of about 0.74. This means that our approach can identify approximately 74% of smell-inducing issues before implementation begins. When design alternatives are still being considered. Our approach provides early warnings of potential architectural risks. This work shifts from reactive remediation to proactive quality assurance, raising awareness of potential architectural risks
Frühlingsbeginn und Mandelblüte 2026 an der Unterhaardt / Début de printemps 2026 et floraisons des amandiers dans la Unterhaardt
No full textDieser Bericht dokumentiert den phänologischen Frühlingsbeginn 2026 in der Unterhaardt-Region (Rheinland-Pfalz, Deutschland) mit Fokus auf die frühe Blüte der Mandelbäume (Prunus amygdalus) in Grünstadt. Die ersten Blüten wurden am 27. Februar 2026 beobachtet, die volle Blüte fotografisch am 3. März 2026 festgehalten – 12 Tage früher als im Vorjahr. Die Studie verzeichnet zudem das Auftreten exotischer Halsbandsittiche (Psittacula krameri) in den blühenden Bäumen, was ihre Etablierung in der lokalen Avifauna verdeutlicht. Diese Beobachtungen ergänzen eine langjährige phänologische Datensammlung (2015–2026) und unterstreichen die Bedeutung von Prunus amygdalus als Bioindikator für Klimawandel in gemäßigten Weinbaugebieten
Optimal Configurations for Modular Systems
No full textThis work investigates how to determine optimal configurations for modular systems, a concept widely used in mechanical engineering to improve logistics and manufacturing processes. Although modular systems are common in industrial applications, systematically deriving an optimal configuration with methods of mathematical optimization remains an open challenge.
After introducing modular systems and reviewing relevant literature, we develop a mathematical optimization model that captures the structural relationships between components, variants, and the number of pieces in the modular system. Because the number of variants for the components is itself a decision variable, the initial formulation leads to a non implementable optimization model. To address this, we derive in a next step a well-defined mixed-integer optimization problem (MIP) that deals with the unknown number of variants. Although state-of-the-art MIP solvers such as Gurobi can solve MIPs with a large number of variables efficiently, numerical experiments reveal high computational effort for the modular system instances considered here, due to their strong combinatorial and logical structure.
To exploit this structure, we propose decomposition-based solution methods and analyze their numerical behavior. Motivated by these results, we introduce discrete functions as an alternative representation and examine discrete convexity concepts, including one newly developed in this work. Building on this, we adapt several derivative-free optimization techniques, such as Steepest Descent and Coordinate Search, and an adaption of the Nelder–Mead method to discrete functions
Amtliche Bekanntmachung. Bekanntmachungen. 2026,3: Satzung für das hochschuleigene Zugangs- und Auswahlverfahren im Masterstudiengang Computational Modeling in Engineering am Karlsruher Institut für Technologie (KIT) vom 26.02.2026
No full textAmtliche Bekanntmachung. Bekanntmachungen. 2026,2: Satzung für den Zugang zu dem Masterstudiengang Umwelt- und Georessourcenmanagement am Karlsruher Institut für Technologie (KIT) vom 26.02.2026
No full textThe surrogate Gibbs-posterior of a corrected stochastic MALA: Towards uncertainty quantification for neural networks
No full textTowards a domain specific graph query language for the geoscience - implementing a GeoGQL -
No full textModern geo-data management plays a crucial role in designing digital twins in distributed system environments, enabling seamless integration, analysis, and visualization of spatial information. With the rise of graph databases and linked data, geospatial relationships can be efficiently modeled and queried using technologies such as Gremlin, a graph traversal language. On the other hand, Simple Features (see ISO19107) and the Dimensionally Extended 9-Intersection Model (DE-9IM) as two traditional examples provide standardized frameworks for spatial representation and topological reasoning, ensuring interoperability across systems. The fusion of geospatial standards with schema-free geo-data management advances the support of real-time decision-making and scalable geospatial applications, making modern geo-data management a cornerstone of intelligent, interconnected digital environments. Giving meaning to standards by ontologies is one major rapprochement to establish semantic interoperability. This paper provides one step towards this goal by using an abstract graph schema to represent the intra- and inter-relations of simplicial- and polytope-complexes and applying the traditional geoinformatics interpretation of topology, the philosophy of the Dimensionally Extended 9-Intersection Model (DE-9IM) to give meaning. This approach can be seen as a step towards the implementation of a Domain-Specific-Language (DSL) for property graphs that represent complex interrelated vector data within a linked data world
Framework for Natural Language Processing to Automate Material Flow Simulation in Production System Planning
No full textAn increasingly dynamic market environment is enforcing greater changeability and flexibility in production systems. To address this complexity with simulation models, expert knowledge needs to be externalized. Therefore, we propose a framework that converts expert discussions and written documents into textual requirements to create simulation configurations. This automates the manual and time-consuming process of specifying the scenario and instantiating simulation models. We discuss the overall architecture that is able to understand complex technical language and explain the approach that dynamically adapts to new requirements. The natural language processing-based framework promises great potential of a seamless setup up of a-priori evaluations of productions systems to enable a comprehensive deployment of digital twins
Development of Low Fluorinated, Sustainable, and Recyclable Electrolytes Based on γ‐Valerolactone for High‐Performance Sodium‐Ion Batteries
No full textThis study introduces a dual salt novel electrolyte for sodium-ion batteries (SIBs), consisting of sodium difluoro(oxalato)borate (NaDFOB) and sodium bis(fluorosulfonyl)imide (NaFSI) salts dissolved in the bio-based γ-valerolactone (GVL) solvent. Besides its renewable origin, the electrolyte exhibited strong inhibition of anodic dissolution and excellent electrochemical stability (up to 4.3 V vs. Na/Na). It delivered outstanding cycling stability, with ∼87 % capacity retention after 100 cycles in P2-NaAlFeMnNiO (P2-AFMNO) cathode half cells and ∼80 % retention after 200 cycles in lab- scale full cells with hard carbon anodes when cycled within a wide voltage window of 1.5–4.3 V. Post mortem X-ray photoelectron spectroscopy analysis helped gaining deeper understanding about the decomposition products formed on the interphases. A simple and sustainable water-based process is employed to successfully recover the GVL solvent. The recovery method enabled recover 85 % of GVL solvent from the recycling process. The feasibility of recycling is further demonstrated by reusing the recovered GVL-based electrolyte in full cells, which achieved performance comparable to that of the pristine GVL-based electrolyte and exhibited excellent long-term stability, retaining approximately 83 % of its capacity after 100 cycles