TY - CONF A1 - Winkler, Nicolas P. A1 - Neumann, Patrick P. A1 - Schaffernicht, E. A1 - Lilienthal, A. J. T1 - Heterogeneous Sensor Networks: Challenges and Insights from an Industrial Scenario N2 - Monitoring airborne pollutants is critical for occupational health, particularly in industrial environments where workers are exposed to hazardous emissions. Traditional measurements are typically limited to single-day campaigns, resulting in extremely sparse temporal data. Low-cost sensor networks offer a way to increase spatial and temporal resolution but are limited by issues of accuracy and reliability. To address this, we present a wireless heterogeneous sensor network that integrates low-cost stationary nodes with high-quality sensors on mobile platforms, including ground and aerial robots. We deploy this system in a hot rolling mill facility and evaluate its performance under real-world conditions. Field experiments reveal dynamic pollutant patterns, such as altitude-dependent PM2.5 gradients and temperature fluctuations. By introducing synchronized “rendezvous” events between mobile and stationary nodes, we demonstrate correlation capabilities of sensors. Our spatiotemporal analysis shows that, despite limitations of mobile sensing, strategically combining heterogeneous data sources enables capturing pollutant dynamics in complex industrial settings. T2 - IEEE SENSORS 2025 CY - Vancouver, BC, Kanada DA - 19.10.2025 KW - Environmental monitoring KW - Sensor data fusion KW - Sensor system networks KW - Mobile robotics PY - 2025 SN - 979-8-3315-4467-6 SP - 1 EP - 4 PB - IEEE CY - Piscataway, NJ, USA AN - OPUS4-64501 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - CONF A1 - Winkler, Nicolas P. T1 - Heterogeneous Sensor Networks: Challenges and Insights from an Industrial Scenario N2 - Monitoring airborne pollutants is critical for occupational health, particularly in industrial environments where workers are exposed to hazardous emissions. Traditional measurements are typically limited to single-day campaigns, resulting in extremely sparse temporal data. Low-cost sensor networks offer a way to increase spatial and temporal resolution but are limited by issues of accuracy and reliability. To address this, we present a wireless heterogeneous sensor network that integrates low-cost stationary nodes with high-quality sensors on mobile platforms, including ground and aerial robots. We deploy this system in a hot rolling mill facility and evaluate its performance under real-world conditions. Field experiments reveal dynamic pollutant patterns, such as altitude-dependent PM2.5 gradients and temperature fluctuations. By introducing synchronized “rendezvous” events between mobile and stationary nodes, we demonstrate correlation capabilities of sensors. Our spatiotemporal analysis shows that, despite limitations of mobile sensing, strategically combining heterogeneous data sources enables capturing pollutant dynamics in complex industrial settings. T2 - IEEE SENSORS 2025 CY - Vancouver, BC, Kanada DA - 19.10.2025 KW - Environmental monitoring KW - Sensor data fusion KW - Sensor system networks PY - 2025 AN - OPUS4-64506 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - CONF A1 - Neumann, Patrick P. A1 - Winkler, Nicolas P. A1 - Nerger, Tino A1 - Lohrke, Heiko A1 - Bartholmai, Matthias T1 - Robotic Olfaction in Action: Field Applications and Results from Current Research N2 - In recent decades, robotics, particularly in environmental monitoring, has made significant advances. Robots of various forms and sizes have become essential tools for data collection in environmental research. Mobile Robot Olfaction (MRO) involves mobile robots equipped with gas sensors and requires the integration of multiple disciplines, including signal processing, machine perception, autonomous navigation, and pattern recognition. Common applications of MRO include mapping gas distributions, locating and detecting gas sources, and tracking gas plumes. Aerial Robot Olfaction (ARO) is a specialized branch of MRO that adapts these concepts to aerial robots, addressing the challenges of airborne gas sensing. This presentation highlights recent developments and results from ongoing research projects in MRO and ARO, with a focus on real-world deployment scenarios and the challenges encountered in practice. T2 - Drohnen in der Zerstörungsfreien Prüfung CY - Magdeburg, Germany DA - 26.11.2025 KW - Ground and Aerial robots KW - Gas distribution mapping KW - Gas source localization KW - Gas Tomography KW - Mobile Robotic Olfaction PY - 2025 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-648928 UR - https://www.ndt.net SP - 1 EP - 15 PB - DGZfP AN - OPUS4-64892 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Winkler, Nicolas P. A1 - Neumann, Patrick P. A1 - Albizu, Natalia A1 - Schaffernicht, Erik A1 - Lilienthal, Achim J. T1 - GNN-DM: A Graph Neural Network Framework for Real-World Gas Distribution Mapping N2 - Gas distribution mapping (GDM) is essential for industrial safety and environmental monitoring, as it enables real-time hazard detection and air quality assessment. Traditional GDM methods, such as kernel-based techniques, struggle to reconstruct complex gas plume dynamics accurately. While deep learning has shown promise for GDM, two critical challenges hinder its practical use: the scarcity of available training data and the incompatibility of conventional architectures with irregular sensor layouts. To address these limitations, we propose GNN-DM, a graph neural network-based model for GDM that incorporates the relational structure of sensor networks to infer high-resolution maps from minimal, irregular inputs. The model is pretrained on synthetic gas dispersion data generated from measured wind data and fine-tuned on two industrial datasets collected on a ferry car deck and in a hot rolling mill. Compared with established GDM techniques, GNN-DM achieves higher accuracy on synthetic and real-world data, highlighting the potential of graph-based learning for practical gas mapping applications. KW - Environmental monitoring KW - Sensor networks KW - Transfer learning KW - Deep learning PY - 2025 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-647363 DO - https://doi.org/10.1109/JSEN.2025.3617158 SN - 1530-437X VL - 25 IS - 22 SP - 42171 EP - 42179 PB - Institute of Electrical and Electronics Engineers (IEEE) AN - OPUS4-64736 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - JOUR A1 - Winkler, Nicolas P. A1 - Kotlyar, O. A1 - Schaffernicht, E. A1 - Matsukura, H. A1 - Ishida, H. A1 - Neumann, Patrick P. A1 - Lilienthal, A. J. T1 - Super-resolution for Gas Distribution Mapping N2 - Gas Distribution Mapping (GDM) is a valuable tool for monitoring the distribution of gases in a wide range of applications, including environmental monitoring, emergency response, and industrial safety. While GDM is actively researched in the scope of gas-sensitive mobile robots (Mobile Robot Olfaction), there is a potential for broader applications utilizing sensor networks. This study aims to address the lack of deep learning approaches in GDM and explore their potential for improved mapping of gas distributions. In this paper, we introduce Gas Distribution Decoder (GDD), a learning-based GDM method. GDD is a deep neural network for spatial interpolation between sparsely distributed sensor measurements that was trained on an extensive data set of realistic-shaped synthetic gas plumes based on actual airflow measurements. As access to ground truth representations of gas distributions remains a challenge in GDM research, we make our data sets, along with our models, publicly available. We test and compare GDD with state-of-the-art models on synthetic and real-world data. Our findings demonstrate that GDD significantly outperforms existing models, demonstrating a 35% improvement in accuracy on synthetic data when measured using the Root Mean Squared Error over the entire distribution map. Notably, GDD appears to have superior capabilities in reconstructing the edges and characteristic shapes of gas plumes compared to traditional models. These potentials offer new possibilities for more accurate and efficient environmental monitoring, and we hope to inspire other researchers to explore learning-based GDM. KW - Gas distribution mapping KW - Spatial interpolation KW - Sensor networks KW - Deep learning PY - 2024 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-607786 DO - https://doi.org/10.1016/j.snb.2024.136267 SN - 0925-4005 VL - 419 SP - 1 EP - 12 PB - Elsevier B.V. AN - OPUS4-60778 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - CONF A1 - Winkler, Nicolas P. A1 - Neumann, Patrick P. A1 - Schaffernicht, Erik A1 - Lilienthal, Achim T1 - Gas Distribution Mapping With Radius-Based, Bi-directional Graph Neural Networks (RABI-GNN) N2 - Gas Distribution Mapping (GDM) is essential in monitoring hazardous environments, where uneven sampling and spatial sparsity of data present significant challenges. Traditional methods for GDM often fall short in accuracy and expressiveness. Modern learning-based approaches employing Convolutional Neural Networks (CNNs) require regular-sized input data, limiting their adaptability to irregular and sparse datasets typically encountered in GDM. This study addresses these shortcomings by showcasing Graph Neural Networks (GNNs) for learningbased GDM on irregular and spatially sparse sensor data. Our Radius-Based, Bi-Directionally connected GNN (RABI-GNN) was trained on a synthetic gas distribution dataset on which it outperforms our previous CNN-based model while overcoming its constraints. We demonstrate the flexibility of RABI-GNN by applying it to real-world data obtained in an industrial steel factory, highlighting promising opportunities for more accurate GDM models. T2 - International Symposium on Olfaction and Electronic Nose (ISOEN) CY - Grapevine, TX, USA DA - 12.05.2024 KW - Gas distribution mapping KW - Spatial interpolation KW - Graph neural networks KW - Mobile robot olfaction PY - 2024 SN - 979-8-3503-7053-9 DO - https://doi.org/10.1109/isoen61239.2024.10556309 SP - 1 EP - 3 PB - IEEE AN - OPUS4-60103 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - CONF A1 - Neumann, Patrick P. A1 - Hüllmann, Dino A1 - Winkler, Nicolas P. A1 - Lohrke, Heiko A1 - Lilienthal, A. J. ED - Lee, J. B. T1 - Outdoor Gas Plume Reconstructions: A Field Study with Aerial Tomography N2 - This paper outlines significant advancements in our previously developed aerial gas tomography system, now optimized to reconstruct 2D tomographic slices of gas plumes with enhanced precision in outdoor environments. The core of our system is an aerial robot equipped with a custom-built 3-axis aerial gimbal, a Tunable Diode Laser Absorption Spectroscopy (TDLAS) sensor for CH4 measurements, a laser rangefinder, and a wide-angle camera, combined with a state-of-the-art gas tomography algorithm. In real-world experiments, we sent the aerial robot along gate-shaped flight patterns over a semi-controlled environment with a static-like gas plume, providing a welldefined ground truth for system evaluation. The reconstructed cross-sectional 2D images closely matched the known ground truth concentration, confirming the system’s high accuracy and reliability. The demonstrated system’s capabilities open doors for potential applications in environmental monitoring and industrial safety, though further testing is planned to ascertain the system’s operational boundaries fully. T2 - 20th International Symposium on Olfaction and Electronic Nose CY - Grapevine, Texas, USA DA - 12.05.2024 KW - Aerial robot KW - TDLAS KW - Gas Tomography KW - Plume PY - 2024 SN - 979-8-3503-4865-1 DO - https://doi.org/10.1109/isoen61239.2024.10556071 SP - 1 EP - 3 PB - IEEE CY - USA AN - OPUS4-60107 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - CONF A1 - Neumann, Patrick P. A1 - Winkler, Nicolas P. A1 - Nerger, Tino A1 - Lohrke, Heiko A1 - Stanisavljevi, Mila T1 - DLR Research Seminar - ARO Lab@BAM – Current Research Topics N2 - This seminar presents the key research activities of ARO Lab@BAM, focusing on five main areas: • Learning-based Gas Distribution Mapping utilizes machine learning to accurately model and predict spatial gas concentrations, enhancing environmental monitoring and safety. • Mimose-A develops autonomous systems using artificial intelligence to enable the early detection of leaks in industrial environments. • AGATO (Gastomography) introduces a novel robotic system for high-resolution gas distribution mapping. • Passive Smart Dust detects chemically hazardous substances using drones equipped to distribute and detect particles carrying selective dyes, enabling rapid and reliable monitoring without complex components. • HyAirLogic Lab advances hydrogen (H₂) research by testing the entire value chain in various Berlin-Brandenburg quarters, addressing technological challenges, public acceptance, and sustainable energy solutions for H₂-cargo drones. T2 - DLR Research Seminar CY - Weßling, Germany DA - 04.11.2024 KW - Aerial robot KW - Learning Based Gas Distribution Mapping KW - Mobile Robotic Olfaction KW - Aerial-based Gas Tomography KW - Passive Smart Dust PY - 2024 AN - OPUS4-61611 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - CONF A1 - Winkler, Nicolas P. T1 - Gas Distribution Mapping With Radius-Based, Bi-directional Graph Neural Networks (RABI-GNN) N2 - Gas Distribution Mapping (GDM) is essential in monitoring hazardous environments, where uneven sampling and spatial sparsity of data present significant challenges. Traditional methods for GDM often fall short in accuracy and expressiveness. Modern learning-based approaches employing Convolutional Neural Networks (CNNs) require regular-sized input data, limiting their adaptability to irregular and sparse datasets typically encountered in GDM. This study addresses these shortcomings by showcasing Graph Neural Networks (GNNs) for learningbased GDM on irregular and spatially sparse sensor data. Our Radius-Based, Bi-Directionally connected GNN (RABI-GNN) was trained on a synthetic gas distribution dataset on which it outperforms our previous CNN-based model while overcoming its constraints. We demonstrate the flexibility of RABI-GNN by applying it to real-world data obtained in an industrial steel factory, highlighting promising opportunities for more accurate GDM models. T2 - International Symposium on Olfaction and Electronic Nose (ISOEN) CY - Grapevine, TX, USA DA - 12.05.2024 KW - Gas distribution mapping KW - Spatial interpolation KW - Graph neural networks KW - Mobile robot olfaction PY - 2024 AN - OPUS4-60106 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - CONF A1 - Winkler, Nicolas P. A1 - Johann, Sergej A1 - Kohlhoff, Harald A1 - Neumann, Patrick P. T1 - Revisiting Environmental Sensing Nodes: Lessons Learned and Way Forward N2 - Setting up sensors for the purpose of environmental monitoring should be a matter of days, but often drags over weeks or even months, preventing scientists from doing real research. Additionally, the newly developed hardware and software solutions are often reinventing existing wheels. In this short paper, we revisit the design of our environmental sensing node that has been monitoring industrial areas over a span of two years. We share our findings and lessons learned. Based on this, we outline how a new generation of sensing node(s) can look like. T2 - SMSI 2023 Conference Sensor and Measurement Science International CY - Nuremberg, Germany DA - 08.05.2023 KW - Sensing node KW - Sensor network KW - Environmental monitoring KW - Low-cost KW - LoRaWAN PY - 2023 SN - 978-3-9819376-8-8 DO - https://doi.org/10.5162/SMSI2023/C5.1 SP - 173 EP - 174 PB - AMA Service GmbH CY - Wunstorf AN - OPUS4-57454 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER -