149544 research outputs found
Sort by
Enhancing Operations at Col-CC by Utilizing LLMs, KGs, and RAG
No full textThis poster presents a hybrid system, which combines Large Language Models (LLMs) with Knowledge Graphs (KGs) and Retrieval-Augmented Generation (RAG) to enhance the operational efficiency of the flight control team at the Columbus Control-Center (Col-CC). Col-CC is responsible for the operations of the Columbus module of the International Space Station (ISS), and is part of German Aerospace Center's (DLR e.V.) German Space Operations Center (GSOC). LLMs have demonstrated a remarkable capacity to comprehend and produce human-like text, positioning themselves as an effective and efficient solution for automating routine tasks and delivering real-time support. However, their effectiveness can be constrained by a lack of domain-specific knowledge and the need for accurate, up-to-date information. To address these limitations, we propose a combination of LLMs with KGs and RAG. KGs offer a structured representation of domain-specific information, enabling more effective access to and utilization of specialized knowledge, while RAG enhances LLMs by retrieving relevant documents and data snippets, ensuring that the generated responses are grounded in current information. By leveraging the strengths of LLMs, KGs, and RAG, this approach aims to create a more intelligent and responsive support system for space missions, ultimately contributing to the safety and success of ISS and Columbus operations
Ensuring maritime safety in the autonomous shipping era: the need for multi-system radionavigation receivers
No full textThe maritime shipping domain plays a vital role in the world economy, accounting for over 90% of global trade activity. As a result, the industry's continuous evolution is essential to sustaining growth worldwide, with the United Nations predicting an average annual growth rate of 2.4% through 2029 for maritime global trades. To achieve this growth, digital technologies and automation will be critical components, enabling the safe operation of Maritime Autonomous Surface Ships (MASS) with varying levels of autonomy.
MASS will heavily rely on highly integrated technological solutions, including reliable and resilient positioning information. Global Satellite Navigation Systems (GNSS) have been increasingly used in the maritime domain to provide accurate position, navigation, and timing (PNT) data. However, the widespread adoption of GNSS has also introduced significant risks, particularly when it comes to intentional GNSS signal interference.
GNSS receivers can be severely compromised by various types of man-made interference, including jamming and spoofing attacks. Jamming involves intentionally injecting noise into the receiver, while spoofing refers to the deliberate transmission of misleading GNSS signals from another source that causes faulty positioning. These threats are noticeably increasing, particularly in areas affected by geopolitical instability.
In the Baltic Sea, for example, jamming is a significant concern which affects the maritime domain as well as the aviation one, especially on its eastern side. To address this issue, regional terrestrial navigation systems like R-Mode have been developed as backup to GNSS. R-Mode provides alternative positioning information and can support the detection of misleading PNT data derived from spoofing attacks, enabling decision-making support for the crew or the autonomous navigation system.
In this contribution the R-Mode system will be presented in details. The main features of the system will be described and key performance indicators, based on predictions and real-world measurement campaign will be shown. Our work in the direction of reliable and resilient provision of PNT data will also be presented, including the concept of multi-sensor PNT unit.
Multi-sensor PNT units will become essential components in the maritime domain. These systems will enable the reliable and resilient provision of positioning data, supporting increased autonomy in the maritime landscape. By leveraging regional terrestrial navigation systems like R-Mode, the maritime industry can mitigate the risks associated with unreliable positioning information. This will enable the safe operation of MASS, support increased autonomy, and ensure the continued growth of maritime industry
Deep-Learning-based Dent Detection of Aircraft Surfaces using Synthetic Data
No full textDetecting and localizing dents on aircraft surfaces is crucial for maintaining their structural integrity. However, this task can be challenging for humans as dents are not very prominent to the naked eye and require the assistance of light reflections to reveal the damages across the surface. Latest state-of-the-art technologies such as lasers or cameras digitize this step, however the work load is shifted in identifying the dents to the virtual image. The integration of deep-learning methodologies can help automate dent detection. This study compares two object detection architectures: You Look Only Once (YOLOv11) and Real-Time DETection TRansformer (RT-DETR) for dent detection. A high quality dent dataset is prepared, consisting of real and synthetic images of common long- and mid-range aircraft fuselages, to train and test the models. The results indicate that YOLOv11 marginally outperforms RT-DETR in detecting dents with a mean accuracy precision (mAP50) score of 0.66 against the mAP50 value of 0.57 for RT-DETR
IMoGer - Innovative Modulare Mobilität made in Germany: Modulare, automatisierte Mobilität mit Anknüpfungspotenzial für mehr
No full textGNSS/IMU seonsor fusion integration framework
No full textGlobal Navigation Satellite Systems (GNSS) have become increasingly widespread, with more satellites being launched and better signal availability than ever before. Due to their convenience and accessibility, GNSS technologies are now being used in a wide range of applications, among them, autonomous navigation for vehicles such as cars, ships, and drones. High accuracy and real-time positioning is crucial for these applications which are rapidly growing.
Despite advancements in satellite technologies, GNSS alone still faces inherent limitations: the update rate is relatively low, and signal blockage or multipath effects in urban environments can hinder continuous and reliable positioning. In contrast, inertial sensors are self-contained and provide high-frequency motion data, making them ideal for capturing rapid dynamics. However, they suffer from drift over time due to the accumulation of measurement errors.
To overcome these limitations, this thesis presents an online oriented GNSS and Inertial Navigation System (INS) integration system that directly processes GNSS and IMU raw measureemnts from the receiver. A quaternion-based Error State Kalman Filter (ESKF) is implemented to perform sensor fusion, combining the advantages of both systems while maintaining numerical stability.
The proposed framework includes a complete data extraction pipeline capable of decoding GNSS and IMU measurements directly from the receivers binary messages, enabling online processing without relying on RINEX files. Experimental validation is carried out using simulated datasets, and the performance is analyzed under various scenarios, including temporary GNSS outages, to assess the robustness of the developed system
Benchmark Laminate Datasets and published results of LP2SS Software (Version 1.0.0)
No full textLaminate Datasets to benchmark performance of layup design methods from lamination parameters. This repository also contains the result of the benchmarking of state-of-the-art solutions and a new in-house developed DLR tool LP2SS. The research paper explaining the novel methods behind LP2SS can be found in this link (https://doi.org/10.1016/j.compstruct.2025.119939)
Semantic Cloud Segmentation Models for Solar Applications with Synthetic Data
No full textHighly resolved intra-hour solar irradiance forecasts benefit the solar industry by boosting plant
efficiency, storage integration, energy trading, and grid stability. These forecasts typically rely
on ground-based all-sky imagers (ASIs) and irradiance measurements. Semantic cloud
segmentation is key for physics-based and certain data-driven ASI forecasting methods.
Despite advances in computer vision, semantically segmenting sky images into distinct cloud
classes remains challenging. Similar visual traits and unclear boundaries in certain cloud types
complicate segmentation, especially in overlapping multi-layer scenarios. Distinguishing thin
high-altitude clouds from atmospheric turbidity and aerosols is difficult due to fuzzy edges.
Variability in cloud appearance, caused by shifting light, atmospheric effects, and fish-eye lens
distortion, adds further complexity. Recent studies [1] classify clouds into low-, mid-, and highlayer categories, plus clear sky, based on WMO definitions [2]. Though simplified, this
approach highlights how cloud composition affects solar irradiance. High-layer clouds (e.g.,
Cirrostratus, Cirrocumulus), composed of ice, typically reduce irradiance slightly. Dense lowlayer clouds (e.g., Cumulus, Stratus), primarily water, dim it significantly. Mid-layer clouds (e.g.,
Altocumulus, Altostratus), blending traits of both, have a broad variety. Individual cloud layers
show distinct dynamics such as movement, speed, development, or dissipation, driven by
tropospheric wind and atmospheric conditions.
Reliable deep learning models for semantic cloud segmentation require extensive, highquality pixel-level annotations. However, manual annotation is costly and impractical for large
datasets. Recent automated methods blending self-supervised and weakly-supervised
techniques with limited manual ground truth data show promise [1, 3]. This work advances
automated annotation by exploiting temporal relationships in sequential images through
semi-supervised video object segmentation, leveraging distinct cloud layer dynamics. The
application of video object segmentation to clouds has not been explored yet highlighting the
novelty of this approach.
[1]: https://doi.org/10.5194/amt-15-797-2022
[2]: https://cloudatlas.wmo.int/en/home.html
[3]: https://elib.dlr.de/204657
Stress during sleep due to sleep restriction: HRV as a stress marker
No full textThis study investigates the influence of sleep restriction and following recovery sleep on heart rate variability (HRV) as an indicator of autonomic cardiac stress. Nineteen healthy young female and male adults (aged 20–26) completed a 12-day, controlled laboratory study, including baseline sleep, a five-night sleep restriction phase, and recovery sleep nights, with polysomnography and ECG-based HRV measurements. HRV markers were calculated for light sleep (N2), deep sleep (N3), rapid eye movement (REM), and five-hour sleep segments. Statistical analysis with mixed ANOVA revealed significant reductions in parasympathetic HRV markers (SDNN, RMSSD, pNN50) and increased heart rate in the last sleep restricted night, with incomplete normalization during recovery sleep. REM and N3 sleep stages showed particular vulnerability to autonomic dysregulation, highlighting limited cardiovascular resilience to sleep restriction. Individual differences in HRV reactivity emphasize the requirement of a personalized approach to assessment and intervention. The results highlight the importance of sleep in preserving autonomic and CV health and suggest the need for prospective longitudinal studies with more prolonged recovery periods to examine their longterm risks/recovery
