Research

The IN+SIGHT project simplifies the creation of immersive training environments through 3D digital twins and Spatial AI, making XR-based learning accessible across sectors such as healthcare, emergency services, public infrastructure, cultural heritage, agriculture, and the green transition. By enabling domain experts to independently capture real-world scenarios using smartphones, enrich them with spatially anchored verbal annotations, and convert them into photorealistic 3D models via 3D Gaussian Splatting (3DGS), IN+SIGHT reduces the complexity and cost of XR training development without requiring specialized 3D skills. IN+SIGHT’s approach consists of three key steps: 1. Capturing: Domain experts use smartphones to capture real-world environments and add verbal annotations, creating the foundation for photorealistic 3D models. 2. Authoring: With intuitive editing tools, domain experts enrich the captured data by referencing relevant documents and define learning paths, supported by Conversational AI. 3. Learning: Through Web 4.0-based technical foundations, trainees access the virtual learning scenarios across multiple platforms – mobile devices, PCs, or immersive headsets – ensuring accessibility and flexibility for a broad range of users. Spatial AI gives contextual guidance based on the location within the digital twin. The project employs user-centered design with feedback from evaluations together with use case partners involved in professional training across a wide range of industries and sectors: ÖBB, Samariterbund Linz, Österreichisches Rotes Kreuz NÖ, Bundesverband Rettungsdienst, Samariterbund Österreich, BOKU University, Verein Energiewende, and Belvedere. The R&D partners St. Pölten University of Applied Sciences, AIT, Mopius, and Johannes Ambrosch (Startup) contribute expertise in XR technologies, AI, and software development. IN+SIGHT’s main benefit is allowing organizations to independently produce & continually update customized XR training environments and to roll them out to a broad set of trainees. This is particularly valuable for institutions with significant societal impact. Results and demonstrators will be shared through public events, publications, and a hackathon to encourage broad adoption.

Intensive land use is contributing to habitat loss and ecosystem degradation, which threaten global biodiversity, particularly in grasslands that are home to diverse plant and insect communities. Most past research has focused on the direct effects of land use on insects, but indirect effects, such as changes in microclimate, have not been fully explored. How land use affects microclimatic niche space, and how this relationship moderates the negative impact of land-use intensity on insect diversity, is unknown. We will combine spatially highly-resolved microclimate data measured by thermal remote sensing with insect sampling and the measurement of thermal tolerances of insects to understand how land-use driven changes in microclimate relate to insect diversity and community assembly. We hypothesize that intensively-used grasslands will provide more narrow microclimatic niches and that different land-use components (mowing, fertilization, grazing), relate differently to microclimate. To comprehensively test this hypothesis, we will quantify the occurrence and diversity of insects in all 150 grassland plots with suction sampling (biocoenometer). At the same time, we will repeatedly survey surface temperatures at a high spatial resolution with drone flights to characterize microclimatic niches and to match insect samples with precise microclimate data. Using these surface temperatures and a range of covariates (e.g. plant cover, meteorological conditions) we aim to spatially characterize microclimate within and below the grassland canopy, as perceived by insects. To link thermal tolerance and resilience of insects to the microclimatic conditions they are exposed to, we will measure critical thermal maxima for a subset of insect species. These comprehensive and complementary data from remote sensing and insect ecology will help to disentangle the relationship between land use, microclimate and insect biodiversity in managed grasslands. Hence, the proposed research will considerably advance the understanding how different forms and intensities of land use affect biodiversity, which is one of the central questions of the entire Biodiversity Exploratories programme.

This project involves the development of advanced machine learning (ML) frameworks based on transfer learning that integrate multi-sensor remote sensing data. These frameworks support detailed land cover mapping and the assessment of agricultural plant health. The approach leverages pre-trained ML models and adapts them systematically across different agro-ecological regions, management systems and sensor configurations. Specific consideration is given to high-spatial-and-temporal-resolution data from the Sentinel-2 and Planet constellations. The project aims to improve the robustness and generalisation capacity of ML models under varying environmental conditions by combining complementary spectral, spatial and temporal information. There is a particular focus on transferability across sites and seasons to reduce the need for extensive local training data while maintaining high classification and diagnostic accuracy. The resulting frameworks are designed to be scalable and operational, enabling the consistent monitoring of vegetation status, crop conditions and land cover dynamics. They will support decision-making in precision agriculture, ecosystem monitoring and sustainable land management.