Search

Remote Sensing-based change detection (CD) focuses on the identification of Earth's surface transformations through analysis of multi-temporal satellite images captured for the same geographic region at different points in time. Two common classification-based change detection techniques are post-classification comparison of land cover maps and direct classification of image differences between two periods of interest. In either approach, the selection of an appropriate classifier is critical. Consequently, this study focuses on assessing the performance of different classifiers, including the multi-layer perceptron neural network (MLPNN), support vector machine (SVM), random forest (RF), maximum likelihood (MLH), k-nearest neighbor (KNN), and gaussian naïve bayes (GNB). For this evaluation, two high resolution images captured in 2016 and 2020 by the PS2 sensor within the PlanetScope satellite constellation were used. Additionally, a novel unbiased sampling technique was introduced to selectively capture a minimal number of reference pixels. In the context of change detection, slight variations were observed in classification performance rankings between post-classification and difference image classification methods. However, a consistent trend emerged. The MLPNN consistently achieved the highest accuracy, closely followed by RF and SVM as the second or third-best performers in each technique. In contrast, GNB consistently yielded less favourable results. Importantly, our findings highlight the persistent superiority of the difference image classification in terms of change detection accuracy across all six classifiers. Furthermore, this method offers a significant advantage due to its reduced processing time and computational demands, positioning it as the preferred choice for binary change detection when compared to post-classification techniques.

The present research, conducted at the Monastery of St. Catherine in Ferrara, Italy, is an integral component of a broader investigation focused on comprehending, recognizing and enhancing the tangible values of the Dominican priests and their sacred sites in Ferrara, utilizing advanced digital surveying techniques. The architectural legacy of this monastery serves as a insightful link to the city’s history and inhabitants. Regrettably, its typological features are frequently overlooked or altered, underscoring the need to grasp these characteristics as a fundamental step in ensuring their preservation. Given the monastery’s distinct typological and geometrical attributes, an integrated survey approach was implemented to gather essential data for reconstructing the building in its existing state, encompassing both an urban and detailed scale. This comprehensive assessment sought to evaluate the critical conditions of the construction that required improvement.The objective of this paper is to underscore the advancements achieved in assessing structural health behaviour, incorporating data obtained through terrestrial laser scanning (i.e., TLS) and structure-from-motion techniques, coupled with an unmanned aerial system (i.e., UAS), to generate a three-dimensional data model that accurately captures the entire spatiality of the building and its urban context. As a result, the building assessment demonstrated which parts of the building’s facades were overturning, while identifying the thorough kinematic phenomena progressively acting. The related considerations lead to a structural project perfectly shaped on the unique characteristics of the complex supported by data as accurately as possible, which also positively influenced the entire architectural design process.

The presence of speckle in RADARSAT Constellation Mission (RCM) Compact Polarimetry (CP) Synthetic Aperture Radar (SAR) images can impair the performance of information extraction applications such as classification. Therefore, a critical preprocessing step known as despeckling is necessary to mitigate this granular, noise-like phenomenon in these images. This paper compared several PolSAR speckle reduction methods, including Box Car, IDAN, Lee Refined, Lee Sigma, Improved Lee Sigma, and Lopez filters. A CP SAR dataset collected over agricultural land in southern Quebec, QC, Canada, was utilized for the study. The assessment of despeckling was based on various no-reference quantitative indicators. Each despeckling method was evaluated for its effectiveness in reducing speckle in homogeneous areas, preserving details, and avoiding radiometric distortion. Additionally, the impact of despeckling on the classification of this agricultural land was assessed using the Random Forest classifier. The Stokes parameters, m-chi decomposition, and intensity images were utilized for this purpose. Experimental results indicated that the Box Car method excelled in speckle suppression at the expense of edge over-smoothing. Furthermore, the Lee Sigma and Improved Lee Sigma methods were the most effective in speckle reduction from homogeneous areas while preserving edges and preventing radiometric distortion. Moreover, the classification results demonstrated that appropriate despeckling could significantly enhance classification accuracy.

Integrating optical images with Light Detection and Ranging (LiDAR) data is an important advance in Photogrammetry, Geomatics and Computer Vision, registering the strengths of both modalities (height and spectral information). Most orthoimages and aerial LiDAR data are georeferenced to a common ground coordinate system; however, a registration gap remains, and achieving high-accuracy registration between these datasets is challenging due to their differing data formats and frames of reference. In this paper, we propose an approach to enhance camera-LiDAR registration through combined LiDAR feature layer generation and Deep Learning. Our method involves creating weighted combinations of feature layers from LiDAR data, leveraging intensity, elevation, and bearing angle attributes. Subsequently, a 2D-2D Graph Neural Network (GNN) pipeline serves as an intermediate step for feature detection and matching, followed by a 2D-3D affine transformation model to register optical images to point clouds. Experimental validation across aerial scenes demonstrates significant improvements in registration accuracy. Notably, in urban building areas, we achieved an RMSE of around 1.1 pixel, marking a reduction of 5 pixels compared to georeferenced baseline values. In rural road scenes, our method yielded a pixel RMSE of 1.3, with a 4-pixel reduction compared to baseline results. Additionally, in water scenes, which tend to be noisy in LiDAR data, we achieved a pixel RMSE of 1.8, representing a slight half-pixel reduction compared to the baseline. Therefore, by using weighted and combined LiDAR feature layer and GNN feature matching, this approach augments the number of key points and matches, directly correlating with the observed registration reduction in pixel RMSE across diverse aerial scene types.

Aerial and satellite imagery can provide vital information to relief organizations about the extent and distribution of damages after natural disasters. With manual change detection being too inefficient to be effective, the pursuit of automated change detection has accelerated with the recent developments of deep learning methods. Off-nadir imagery (captured not directly overhead) is the fastest to acquire post-disaster, making it ideal for disaster management scenarios. However, the changes in viewing angles result in shadows and occlusions, making damage detection more difficult. Differences in illumination conditions are ever present in bitemporal aerial and satellite imagery, especially for off-nadir imagery, where the reflectance angle affects the amount of light returning to the sensor, making it harder to detect changes and damages. The hypothesis of this study was that artificial intelligence methods fail to adequately account for the illumination differences between images. To test this hypothesis, two radiometric enhancements, matching and equalization, were applied to four change and damage detection datasets, including a damage detection dataset from the 2010 Haiti earthquake. Using a leading high accuracy fusion convolutional neural network architecture called Changer, improvements of up to 20 percent for F1-Score, a popular remote sensing metric for quantifying the number of correctly classified pixels for specific datasets, were achieved through applying radiometric enhancement techniques. Applying radiometric enhancements on a case-by-case basis led to considerable improvements in accuracy, showing the promise of radiometric enhancement. Lower accuracies were achieved on the Haiti dataset, outlining the need for large disaster-specific datasets for training.

Nitrogen is a plant growth limiting nutrient in natural ecosystems, however, many agricultural systems are saturated due to high fertilizer applications. This can lead to higher costs, nitrogen leakage into the environment and unnecessary GHG emissions that contribute to climate change. To avoid this, Neilson (2021) has proposed an approach based on minimal input agricultural systems (MIAS). MIAS seeks to grow crops with limited fertilizer inputs. This is accomplished through targeted/precision fertilizer placement, managing plant physiology to operate at higher efficiencies and adopting new varieties. To work optimally MIAS requires methods to quickly assess plant nutrient status and adjust fertilizer applications accordingly. This study tested remote sensing for detecting stress in potato plants, based on two separate and independent laboratory remote sensing experiments. The goal is to determine minimal input levels applied to a starvation agricultural system that provide yields equivalent (or perhaps even improving upon) those obtained using current excessive inputs, both in terms of yield and, importantly, quality. This research is at the front-end of a proposed paradigm shifting new approach to agriculture. We are being careful to start at first principles in this work; thus the study presented here is based on multiple trials and independent tests. Results testing experimental approaches and N deprivation assessment determined the optimal leaf density and timing of measurements and demonstrated a capability to detect vegetation stress and N deprivation in three potato varieties. These results will inform next steps for future RPAS/UAV/airborne/satellite studies and be used to develop other plant physiology assessment methods.

LoD2 building models can be used in different digital twin-related applications such as urban planning, disaster management, optimizing green energy efficiency, and solar panel recommendation. Existing technology for 3D modelling of buildings still relies on a large amount of manual work due to the irregular geometries of different roof types. Wireframes have shown to be an effective representation for 3D building especially in LoD2 format. Due to the complexity and diversity of roof types in urban areas, 3D building modeling remains a challenging task. In this paper, we propose a new framework for generating 3D wireframes to model different roof types. While high-resolution airborne images can be utilized to exploit the fine details of the roofs, they have difficulties in areas with poor contrast or shadows. The proposed framework incorporates the Digital Surface Model (DSM) as an auxiliary data source to address this limitation. In this work, we focus on the extraction of roof geometrical components including lines and planes of individual buildings to achieve a consistent LoD-2 building reconstruction. The proposed methodology is divided into two phases: (1) jointly predicting building lines and roof planes from the RGB imagery and DSM and (2) generating 3D wireframes of buildings using the extracted roof planes and lines. Subsequently, height values from the point clouds are used to derive 3D wireframes. Experiments with 1,620 buildings from Fredericton, the capital of New Brunswick in eastern Canada, demonstrate an IoU of 0.9337, an F1-score of 0.939, and an F2-score of 0.9378 for the roof geometrical components detection phase, as well as an RMSE of around 0.2-0.8 meter for the final 3D building model compared to the original LiDAR data was achieved.

Evaluation of surface distress is an important aspect of pavement management. The most common practice to assess surface distress is to develop a pavement condition index (PCI), with ASTM-PCI being the most widely used in evaluating flexible pavements. Traditional PCI evaluation methods rely on labour-intensive, manual inspections, leading to significant time consumption. In recent years, real-time visualization and crowdsourcing have been explored, but their potential has yet to be fully realized. Integrating real-time visualization through GIS technology offers immediate insights into pavement conditions, aiding prompt decision-making. Crowdsourcing allows a broader community to contribute to condition reporting, enhancing data accuracy and coverage. This study aims to develop an artificial intelligence (AI)-based method for road condition assessment from crowd-sourced images. A deep-learning object detection model is utilized for precise crack detection and classification. The model is trained to recognize various distress types accurately and quantify attributes crucial for determining the PCI. The developed model is then integrated into a web-based platform accessible through mobile phones and dash cameras, allowing real-time capture and classification of cracks. The study demonstrates that the automated methodology significantly enhances PCI estimation efficiency, with a high correlation between semi-automated and automated methods. Stakeholders can benefit from deep learning and automation in pavement distress detection, aiding informed decision-making through crowdsourcing data. Future work includes the detection of subclasses within crack types based on severity and the creation of digital twins for public assets. Overall, this study highlights the transformative potential of AI and crowdsourcing in improving pavement management.

Tailings, a byproduct of mining, consist of fine sediment particles suspended in water that are stored in tailings storage facilities (TSFs). The discharge of untreated TSF water into the environment is typically prohibited due to its contact with mine tailings and processing chemicals. TSF failures have caused damage to communities and the environment, prompting calls for better management practices and advanced monitoring tools. For operational mine water management, boat-based bathymetric surveys have been used. However, these technologies have limitations, especially when the surveying of large facilities is required. Advances in remote sensing, particularly satellite-based earth observation (SBEO), offer cost-effective solutions for monitoring TSFs. This study explores the use of machine learning models, including XGBoost and Convolutional Neural Networks (CNNs), applied to Sentinel-2 and Landsat-8/9 data to estimate TSF bathymetry. Surveyed bathymetry datasets were used for model training, testing, and results validation. The results of the experiments revealed that high-accuracy bathymetric estimates could be obtained with mean absolute errors between 6 and 12 cm depending on the source of the data (i.e. Sentinel-2 or Landsat-8/9) and the model used (XGBoost vs CNN). Limitations include mixed pixel effects on the pond-beach interface and lower accuracies obtained in shallow areas, notably when XGBoost is used. This research underscores the potential of using satellite data and machine learning for TSF bathymetric monitoring, with implications for enhancing environmental and safety standards in mining operations.

The distribution of tree species within the forest holds significant importance for forest management. Since field surveys in the forest are time-consuming and cost-expensive, automatically extracting tree species distribution maps from remote sensing imagery becomes a trend. For tree species classification using hyperspectral imagery, many existing classification methods require a large number of training samples to achieve high classification accuracy. However, the classification accuracy will decrease rapidly if only a few hundred training samples are used. Given the challenges and expenses associated with collecting abundant training samples in the forest, there is a need to explore methods that achieve good classification performance with a limited number of training samples. In this paper, a classification scheme combining SuperPCA and Active Learning (AL) is proposed to improve the tree species classification using a limited number of training samples. SuperPCA is employed to reduce feature dimensions and harness spectral-spatial information within hyperspectral imagery. Active Learning is employed to select informative samples for the training, thus reducing the requirement for training samples. Experiments on a tree species classification data set demonstrate the effectiveness of the proposed classification scheme.

With the rise in usage of digital twins in the field of urban planning, the integration of data sources such as sensor data in 3D city models is a challenge that is often brought up. This is even more so with movement data, as they contain a geographical or geometrical aspect that is often overlooked. This paper looks at existing methods to integrate sensor data with 3D city models, and especially at the Dynamizer module in CityGML that supports the integration of dynamic data into the models directly. We look at how the Dynamizer module works with movement data and compare it with another method, SensorThings API. The comprehensive investigation we conduct involved different types of movement data, at different spatial and temporal scales, e.g. origin-destination public transport data in Singapore and London, migration routes in the USA, and highly detailed customer movement in a supermarket in China. While the Dynamizer is more flexible in representing data, it has less support and it is more restrictive when accessing the data. On the other hand, the FROST-Server implementation of the SensorThings API is more restrictive in representing data, however, accessing the data is more flexible.

Building energy system management is critical for resource-saving approaches amid climate change-driven energy transitions. This paper presents a digital twin toolchain leveraging modern technologies such as Building Information Modeling (BIM), Artificial Intelligence (AI), Virtual Reality (VR), and Augmented Reality (AR). The toolchain automates the derivation of georeferenced digital twins during Technical Building Equipment (TBE) commissioning. Using a Scan vs. BIM process, discrepancies between as-planned and as-built TBE are identified, allowing automatic updates to the BIM model. Validation methods ensure both physical and functional aspects of the TBE are accurate. VR and AR facilitate off- and on-site commissioning, enabling immersive visualization and live sensor data access. An evaluation in small and large-scale demonstrators shows the toolchain's scalability and efficiency, with promising results in performance and accuracy. Future work aims to integrate more operational data, enhancing the digital twin's capabilities for building energy system management.

The integration of geoinformation with Building Information Models (BIM), termed GeoBIM, has garnered significant attention across academic and non-academic sectors due to its potential for analyzing the reciprocal impacts of new designs on their environment. However, achieving integration between 3D city models and BIM necessitates ensuring consistency and alignment between their respective features and specifications. Georeferencing, a fundamental task in GeoBIM, involves establishing a connection between digital models and the Earth’s surface through coordinate transformations. Despite its importance, accurate georeferencing of BIM models has often been overlooked, resulting in challenges for integrating BIM models and geographical data. To address this gap, our study proposes a novel approach to enhance the georeferencing accuracy of BIM models by integrating surveyed points, considering the varying levels of georeferencing precision applicable to Industry Foundation Classes (IFC) models. We explore the potential benefits and challenges associated with this integrated surveyed point methodology, providing insights to improve georeferencing within the GeoBIM framework.

In recent years, laser scanning and other survey techniques have become popular for documenting and preserving architectural and structural heritage, including masonry bridges. These bridges require careful monitoring to ensure proper structural integrity analyses. However, analysing large datasets of complex point clouds can be challenging and time consuming. This research introduces an automated method to quickly assess the geometric characteristics of vault of bridges by means of a simple algorithm purposely developed for this application. The developed methodology was implemented in MATLAB. The main geometric features, such as length, arrow, and span, are extracted, and conditional algorithms classify the vault types based on specific thresholds. The methodology is tested on surveyed masonry bridges coming from a case study: the historical Circumetnea railway bridges in Sicily, Italy. A digital survey campaign was conducted using various survey techniques such as laser scanning and photogrammetry to obtain the vault point clouds. This research represents a step forward from previous methodologies by efficiently segmenting and analysing vaults from point clouds, enabling accurate classification and geometric analysis.

The use of Unmanned Aerial Vehicles (UAV) has increased in the last decade, using these technologies for many different topics, such as infrastructure inspection or biodiversity monitoring. The main advantages of these systems are its reduced cost and its ability to navigate on hard-accessible areas, what makes it possible to acquire aerial data with a reduced cost. The new technological developments and the progress in the definition of the regulations applicable to these vehicles have allowed the first BVLOS (Beyond Visual Line of Sight) operations to begin to be carried out at a professional level, which has meant a great advance for the UAV sector. In this field, Telespazio Ibérica SLU is leading a project called GEOSUB, which objective is the development of a UAV integral solution for biodiversity monitoring. The project has two main parts. The first one is the evolution and improvement of a fixed-wing and multicopter UAV, equipped with four different sensors: RGB camera, multispectral camera, thermographic camera, and a LiDAR sensor. The second part focused on the development of a geoinformation processing platform for both the mission planification and monitoring, and the data processing to obtain the final product. This work presents the preliminary results of the project and goes in depth with the developments made.

This paper discusses the representation of the scale dimension using tGAP/SSC-DHE, emphasizing its role in creating vario-scale representations. It outlines the process of integrating the scale dimension as the third dimension in a 2D map before transitioning to 3D or 4D maps. DHE is introduced as an effective data structure for this approach, maintaining crucial connections across different levels of detail and offering enhanced topology. Its strength lies in the ability to modify specific parts of the model while retaining a clear structure, thanks to well-defined operators such as Euler operators. This method provides a versatile framework for representing complex spatial data, allowing for detailed analysis while preserving overall coherence and it updates the shape and structure of the model at the same time through local modifications. Through practical implementation, it demonstrates the potential for improved visualization and understanding of multi-dimensional spatial relationships. Also, the DHE data structure supports slicing by either a horizontal plane to produce a map at a specific scale (depending on the height of the horizontal plane) and tilted planes to produce perspective views with mixed scales.

Porches, as defined by the Art & Architecture Thesaurus, serve as vital transitional spaces linking indoor and outdoor environments. Despite their historical and contemporary significance, porches lack explicit representation in prevalent standards like CityGML and IndoorGML, posing challenges for comprehensive spatial modeling and its application. This paper proposes a method for modeling porches that aligns with the existing OGC standard CityGML 3.0, ensuring accuracy and compatibility. Drawing upon geomatics techniques, the method aims to bridge the gap in representing these spaces, critical for applications such as navigation systems, urban planning, and energy simulations. By integrating geometric, machine learning, and informative modeling approaches, this method seeks to provide a robust foundation for various practical applications. The paper outlines a comprehensive state-of-the-art review, describes the proposed method from digitalization to random forest (RF)-based point cloud classification and vectorization, presents case studies and results, and offers critical discussions and conclusions. Through this endeavor, the paper contributes to enhancing the representation and understanding of porches within the digital spatial landscape.

Building Information Modelling (BIM) and Geographic Information System (GIS) integration allows for analysis and visualization of geospatial characteristics and relationships of buildings and their surroundings for in a number of domains such as planning and management of built environments and disaster management. BIM provides detailed geometric and semantic data about a building and its elements. Moreover, it is a technology that enables 3D display of architectural, structural, and mechanical, electrical, and plumbing (MEP) projects in a local cartesian coordinate system. On the other hand, GIS employs global (geographic or projected) coordinate systems, thereby allowing for the integration, analysis, and visualization of various geospatial data. The development of BIM and GIS for different purposes brings about differences in spatial focus, geometric and semantic representations, level of detail, coordinate systems, and other aspects. In order to eliminate these differences and ensure integration, geometric transformation and semantic transfer are required. The semantic web represents a logical preference for semantic transfer, characterised by a natural ability to integrate information from disparate sources. In this study, the detailed geometric and semantic data of a building with structural, architectural, and MEP projects in BIM were integrated with the geospatial data in GIS using a semantic web approach. For this purpose, an ontology including concepts, relationships, and attributes was developed with the contribution of core ontologies from BIM and GIS. Semantic and geometric data values related to the building and its elements were transferred to this developed ontology, which was formatted in RDF format. Queries were made using SPARQL to analyse the transferred data. The results of these queries indicated that the semantic web is well suited for use in BIM and GIS integration.

We introduce CityJSON Text Sequences (CityJSONSeq in short), a format based on CityJSON and JSON Text Sequences. CityJSONSeq was added to the CityJSON specifications version 2.0 to allow us to stream very large 3D city models. The main idea is to decompose a CityJSON dataset into its individual city objects (each building, each tree, etc.) and create several independent JSON objects of a newly defined type: CityJSONFeature. We elaborate on the engineering decisions that were taken to develop CityJSONSeq, we present the open-source software we have developed to convert to and from CityJSONSeq, and we discuss different aspects of the new format, eg filesize, usability, memory footprint, etc. For several use-cases, we consider CityJSONSeq to be a better format than CityJSON because: (1) once serialised it is about 10% more compact; (2) it takes an order of magnitude less time to process; and (3) it uses significantly less memory.

Intertidal macroalgae play a vital role in marine ecosystems, necessitating effective monitoring of their coverage and diversity. Traditional monitoring methods are labour-intensive and costly, prompting exploration of the use of unmanned aerial vehicles (UAVs) to characterize intertidal ecosystems. We propose an alternative process integrating UAV red-green-blue (RGB) imagery and topographic indexes to classify complex intertidal macroalgae assemblages automatically. We studied two intertidal areas capturing eight flights between May and September 2023. Orthoimages and Digital Elevation Models (DEMs) were generated. Manual segmentations for 24 classes were cropped into images of individual labels. Additional channels with five topographic indices were added to the RGB images. The resulting dataset of 6412 images was then used to train a Convolutional Neural Network (CNN). We tested the benefit of the additional topographic indices by training the CNN with and without the topographic channels. The best results were given by the inclusion of the Analytical hillshade to the RGB images, showing a relative 11.3% increase in classification accuracy. This indicates that 3D data can enhance the performance of macroalgae classification models. However, there was no significant improvement when using more than one topographic index to train the CNN. Our workflow offers a cost-effective and robust solution for intertidal macroalgae monitoring, contributing to ecological conservation efforts.

Developing a precise tree stem curve and robust estimation of stem volume are crucial for forest inventories with various applications. Laser scanned point clouds have been recognized as the most practical data for tree information modeling. Many methods for stem curve development involve estimating stem diameters at different heights and determining stem volume by utilizing fitted cylinders based on these diameters and the associated heights. The estimation of diameter depends on circle fitting. However, many circle fitting methods are non-robust and inaccurate in the presence of noise, outliers, and significant data gaps, resulting in faulty diameters and imprecise stem volume. Limited scanning, occlusions from the physical complexity, high tree density, and adjacent branches may cause data incompleteness, and generate outliers. To address these challenges, we employ robust statistical approaches to restrain the influence of outliers and data gaps. This paper contributes by (i) exploring the problems of robust diameter estimation for partial data, and in the presence of noise and outliers, (ii) understanding the impacts of using erroneous diameters in cylinder fitting, and later for stem curve and volume estimation, and (iii) developing a robust method that couples robust circle and cylinder fittings to derive precise stem curve and estimation of stem volume in the presence of outliers and partial data. We demonstrate the performance of the proposed algorithm through terrestrial laser scanning point clouds.

Nowadays, there are many publicly available georeferenced data, like 3D CityGML models, that can be used as prior knowledge to perform accurate global localization. Iterative Closest Point (ICP) is a promising method for achieving this task, but it requires two point clouds that need to be partially overlapping in the initial state for better registration performance. Therefore, we investigated different detection and matching methods to automatically pre-register two non-overlapping point clouds based on a 2D overhead view and evaluated the registration results produced by an ICP algorithm. We used public data from the city of Aachen, Germany. A georeferenced point cloud was derived from the LOD2 CityGML model and a local point cloud was reconstructed from an image sequence using Structure from Motion (SFM). The evaluation results show that georeferenced LOD2 CityGML models can successfully be used for city-scale sub-meter global localization.

Semantic segmentation of 3D urban environments plays an important role in urban planning, management, and analysis. This paper presents an exploration of leveraging BuildingGNN, a deep learning framework for semantic segmentation of 3D building models, and the subsequent conversion of semantic labels into CityGML, the standardized format for 3D city models. The study begins with a methodology outlining the acquisition of a labelled dataset from BuildingNet and the necessary preprocessing steps for compatibility with BuildingGNN's architecture. The training process involves deep learning techniques tailored for 3D building structures, yielding insights into model performance metrics such as Intersection over Union (IoU) for several architectural components. Evaluation of the trained model highlights its accuracy and reliability, albeit with challenges observed, particularly in segmenting certain classes like doors. Moreover, the conversion of semantic labels into CityGML format is discussed, emphasizing the importance of data quality and meticulous annotation practices. The experiment as described in the methodology shows that outputs from the BuildingGNN for semantic segmentation can be utilized for the generation of CityGML building elements with some percentage of success. This particular work reveals several challenges such as the identification of individual architectural elements based on geometry groups. We believe that the improvement of the segmentation process could be further investigated in our near future work.

We develop a Transformer-based model enhanced with a Channel Attention Module (CAM) to capture the inter-channel dependencies in 3D hyperspectral point cloud data for geological applications. We hypothesize that specific channels of hyperspectral data correspond to distinct mineral types, and therefore, exploiting the relationships among these channels is beneficial for our analysis. We evaluate our method using the newly released Tinto dataset, which consists of 3D hyperspectral point clouds featuring three different spectral ranges: LongWave Infrared (LWIR), ShortWave Infrared (SWIR), and Visible-Near Infrared (VNIR).We explore four different CAMs from various networks—SENet, ECANet, CBAM, and DANet—and successfully integrate them into a CNN-based model to enhance feature representation. We specifically tailor the channel attention to our use of 3D hyperspectral point cloud data. Our experiments demonstrate significant improvements in performance after incorporating the CAM into our backbone model, which draws inspiration from the Point Cloud Transformer architecture and Vector Self-Attention mechanism. These results highlight the potential for further research into enhancing classification accuracy using hyperspectral data in geological applications. The code will be released on https://github.com/aldinorizaldy/CAM-Transformer.

Urban air quality poses a significant public health challenge, particularly in complex urban areas like street canyons, where traffic emissions intensify the issue. The limitations of traditional monitoring methods, marked by a narrow scope and poor visualisation, hinder our understanding of pollutant dispersion dynamics, impeding effective air quality management. Consequently, this study explores the synergy of 3D city models and Computational Fluid Dynamics (CFD) simulation to analyse air pollutant dispersion in complex urban settings. To enhance geo-visualisation, the research employs the next generation of Open Geospatial Consortium (OGC) APIs for data delivery. The integration of 3D city models with CFD simulations and geo-visualisation techniques aims to develop an urban digital twin. This digital twin is envisioned to offer valuable insights into the dynamics of air pollutant dispersion within street canyons, thereby informing more effective air quality management strategies.

The potential of Building Information Models (BIM) to establish rights and responsibilities for multi-level building complexes in cadastral registrations has been explored in many previous researches. However, the implementation of BIM-based cadastral registrations in practice remains limited due to the complex interplay between technical potentials and legal implications as well as uncertainty about the additional complexity of a BIM model and the extra work that will be required to generate the BIM Legal model. In collaboration with Netherlands Kadaster, we have investigated the data requirements for a BIM Legal model that will support the 3D cadastral registration of apartment complexes and that aligns with BIM creation processes in practice. The BIM Legal model aims to visualise the spatial aspects of apartment complexes and to align the cadastral registration to the 3D data that is generated in the design and construction phase of new buildings. The BIM Legal file will be submitted when an apartment complex is being registered and is based on IFC, the open standard established by BuildingSMART International and used in the BIM domain. To make the implementation of BIM Legal feasible legally, technically, and economically, we apply a 3-phase approach. This paper presents the BIM Legal model as defined in Phase 1, which enables to generate a BIM Legal model compliant with current legislation frameworks with no/minimal manual interaction from BIM models as commonly generated in design processes. The paper describes the 3-phase approach, the context of Phase 1, the data requirement analyses, the defined BIM Legal data model, as well the questions that emerged during the specification process that need to be answered to further improve the implementation of Phase 1 and to further develop BIM Legal for Phases 2 and 3.

The growing adoption of digital twins in geomatics sectors requires efficient 3D mapping techniques. However, the complexity and cost of producing cartographically enriched 3D scenes pose significant challenges, hindering widespread application, particularly in domains with limited mapping expertise and budgets. The proposed methodology in this paper leverages recent advances in natural language processing and artificial intelligence, particularly large language models (LLMs), to reduce the expertise required for 3D mapping and to address the high costs and complexity associated with traditional cartographic processes. It introduces a natural language interface for 3D symbology, aimed at simplifying the design and automating the creation of cartographically enriched 3D scene. By allowing cartographers converse with the mapping system, the system translates verbal descriptions into structured symbology rules in 3D digital cartographic model, which are then used to generate cartographically enriched 3D scenes. The method chains multiple ad-hoc LLM-based agents for entity linking, conversation handling, and symbology rule verification. Prompt engineering methods, such as chain-of-though and retrieval augmented generation, have been used to guide the agents’ reasoning process or leverage knowledge base, respectively. Experiment and application in utility networks demonstrates the method’s capability to accurately interpret and execute 3D symbology rules from natural language inputs, resulting in cartographically enriched 3D scenes that are reproducible and scalable. This work represents a pioneer study to implement a natural language interface for 3D mapping. It not only enhances the usability and accessibility of 3D mapping in digital twins but also sets a foundational method for future research in natural language-based mapping interfaces.

Integration of HGIS (Historical Geographic Information Systems) and HBIM (Heritage Building Information Modelling) constitutes a high-precision, efficient, and sustainable method for conserving cultural heritage (CH), particularly in instances characterized by large-scale distribution and spanning multiple historical periods, with extensive archival material. HBIM not only aids in the preservation of remaining structures but also facilitates the revelation and reconstruction of invisible elements. The Chinese Eastern Railway (CER), a key industrial and linear cultural heritage site in northeast China, has a variety of metal bridge heritage, many of which are documented and seen solely in historical archives until now. These bridges, designed and constructed by Russians at the outset of the 20th century, serve as crucial components of the CH along the CER while also showcasing the technological advancements and application developments in metal structures along the railway. We propose a method of integrating HGIS/HBIM to conserve the integrity of the heritage and this methodology has been validated through two case studies. A historical database was established for collection and exhibition archives in QGIS®. Concurrently, Autodesk® Revit was employed to reconstruct bridge models, incorporating material information and GPS coordinates based on the Industry Foundation Classes (IFC). Subsequently, we utilized FME® Workbench to bridge the gap between BIM and GIS. This approach promises to enhance the richness and integrity of the CH along the CER, offering a sustainable solution for its preservation and management.

Transport infrastructures (TIs) play a crucial role for facilitating transportation and enhancing mobility. Nevertheless, they face significant risks from forest fires mainly due to climate change. Developing accurate fire risk maps is essential for planning vegetation maintenance, predicting fire-prone areas, and assessing potential damage levels. Risk maps often rely on field data, which are difficult to update promptly for effective management. This paper introduces an approach to identify and continuously monitor road sections at high risk of forest fires using Random Forest (RF) models along with satellite images. Multiple models were trained using different risk predictors, all achieving test accuracies exceeding 79% and good agreement in the kappa coefficient. Additionally, the key predictors in each model were analyzed. As a result, these models offer a dynamic updating mechanism for risk maps over time. The methodology also enables the integration and fusion of multi-source datasets to pinpoint areas with the highest risk associated with TIs and forest fires. These insights facilitate the delineation and timely update of critical TI areas, thereby enhancing overall risk management.

The here presented Virtual Campus project developed a 3D geodata modelling toolchain, and an open source software based technology stack for creating a Digital geoTwin of the University of Cologne (UoC). We also develop and present a reactjs based web application, a flutter and Unity based mobile app with AR features, as well as an UnrealEngine based VR application to access, browse and visualize the geoTwin data and models in the UoC CAVE 3D environment. The 3D models of the Universities Buildings will include indoor models based on 2D floorplans, which will facilitate indoor positioning in the mobile application of the project through geometry matching and simultaneous localization and mapping (SLAM) technology. For distribution and streaming the 3D building models, we utilze the OGC Standard 3D Tiles, for which we present a tool to georeference the glTF 3D models and assets accurately on the globe.

Consecutive high-temperature days, a phenomenon known as heatwaves, are becoming more frequent and intense due to anthro- pogenic climate change. Padua City, characterized by significant urban soil sealing, is particularly vulnerable to these changes and the exacerbation of Urban Heat Island effects. This study integrates Urban Digital Twin technology and Internet of Things concepts within a three-dimensional modelling environment to develop a Nature-Based Solutions scenario simulation tool. This tool is designed to address climate-manmade problems in Padua City. Using sensor-derived air temperature and relative humidity data, our approach provides detailed micro-climate information to identify heat-prone areas in Padua City. According to this in- formation, the first pilot project test of scenario development was selected to assess how best to achieve a cooling effect through the use of green-blue infrastructure in order to combat the heat hazard in Padua City. Furthermore, this study addresses the urgency of developing Nature-Based Solutions in Padua City’s planning to reduce the heat effect during heatwaves.

The environmental performance of urban neighbourhoods is a crucial indicator to measure the sustainability of the built urban form, especially as cities face increased urbanisation and densified development. With new construction contributing significantly to global carbon emissions, there is a need to understand how modifications in the urban form can affect its environmental performance at the scale of the neighbourhood. This research addresses this need by developing a Planning Support System workflow to aid decision-making in urban planning in order to analyse environmental performance, focusing on thermal comfort and solar energy potential. It explores the relationship between urban form and the selected environmental performance indicators while utilising freely available 3D geoinformation data of the Netherlands. It develops simplified workflows using established plug-ins and open-source 3D software that can be replicated by smaller municipalities with resource constraints, thus removing barriers to 3D-data based data-driven decision making for sustainable and informed urban planning.

Solar energy is becoming increasingly important with the transition towards green and sustainable energy. Predicting solar irradiance is one of the core steps to optimise solar energy utilisation when planning and scheduling power grids. Accurate solar irradiance prediction can also help forecast microclimate conditions, enabling the analysis of citizens and planning of optimal intervention strategies for heating or cooling behaviour. This paper discusses a novel approach to computing the solar potential of buildings at the city level with promising scalability using semantic 3D city models. Experiments are conducted at different locations in the Netherlands. We evaluate our results by comparing them to the statistical Dutch data, and CitySim shows huge discrepancies in summer.

This paper presents a method for stereo visual odometry and mapping that integrates VINS-Fusion-based visual odometry estimation with deep learning techniques for camera pose tracking and stereo image matching. Traditional approaches in the VINS-Fusion relied on classical methods for feature extraction and matching, which often resulted in inaccuracies in triangulation-based 3D position estimation. These inaccuracies could be mitigated by incorporating IMU-based position estimation, which yielded more accurate odometry estimates compared to using stereo camera only in three-dimensional space. Consequently, the original VINS-stereo algorithm necessitated a tightly-coupled integration of IMU sensor measurements with estimated visual odometry.To address these challenges, our work proposes replacing the traditional feature extraction method used in VINS-Fusion, the Shi-Tomasi (Good Features to Track) technique, with feature extraction via the SuperPoint deep network. This approach has demonstrated promising experimental results. Additionally, we have applied deep learning models to the matching of feature points that project the same three-dimensional point to pixel coordinates in different images. Instead of using the KLT optical flow algorithm previously employed by VINS-Fusion, our proposed method utilizes SuperGlue, a deep graph neural network for graph matching, to improve image tracking and stereo image matching performance. The performance of the proposed algorithm is evaluated using the publicly available EuRoC dataset, providing a comparison with existing algorithms.

This study investigates the feasibility of directly utilizing 3D indoor point clouds for real-time indoor navigation, particularly to enhance emergency response processes. Traditional indoor navigation research primarily focuses on creating navigation systems from pre-existing indoor models, resulting in a graph representation that simplifies spatial relationships, requires post-processing, and delivers results only afterwards, often overlooking real-time obstacles and complex layouts such as those in modern office floors. This research proposes an original approach by leveraging real-time generated 3D models using HoloLens 2 sensors, which combine RGB images and depth sensor output to create a comprehensive point cloud. The study explores path planning directly within these point clouds without the need for extensive preprocessing or segmentation, aiming to provide immediate navigation support with minimal delay. Utilizing the Rapidly Exploring Random Trees (RRT) algorithm, the research seeks to minimize preprocessing and swiftly visualize navigable paths, evaluating the system's performance in terms of processing time and path viability. This approach addresses the limitations of traditional graph-based methods and the challenges posed by outdated or unavailable indoor models, offering a promising solution for real-time emergency navigation assistance.

Advanced Air Mobility (AAM) is an emerging sector that aims to optimise transportation in urban and interurban environments using Unmanned Aerial Vehicles (UAVs). However, these aircraft are particularly susceptible to adverse weather phenomena, such as turbulence or wind gusts. To ensure the safety of operations, high-resolution real-time atmospheric prediction models, such as those based on Computational Fluid Dynamics (CFD), are required. The development of these systems faces challenges as accurate geometric models of navigation environments are needed to simulate the interaction between the wind and the terrain and infrastructure. These models feature complex geometries, so they cannot be manually crafted, requiring automated processing of geospatial data sources. As the main novelty, this work presents a methodology for the automated modelling of interurban environments for CFD simulations using digital elevation models and georeferenced surface semantic classification data. A practical case study was developed in the outskirts of Vigo (Spain), where we demonstrate the effectiveness of our approach and its integration with the open-source CFD software OpenFOAM.

Transportation infrastructure serves as the backbone of societal functionality. As aging road networks present increasing challenges, effective monitoring and maintenance become imperative. In this context, this article explores the integration of non-destructive techniques for monitoring and assessing damaged road sections. The work emphasizes the importance of leveraging the complementarity between non-destructive techniques to gain a holistic understanding of damaged road sections' structural integrity. Through the combination of these techniques, structural issues and underlying causes can be identified, as well as to develop maintenance and rehabilitation strategies.A case study of a critical section of the national road N-541 in Pontevedra, Spain, illustrates the application of this integration. First, Interferometry Synthetic Aperture Radar (InSAR) technology revealed points of interest. Four different cases, with two different methods: Scatterer Interferometry (PSI) and Quasi-PS (QPS) were used, to minimize uncertainty. Then, Global Positioning System (GPS) data, together with Laser Imaging Detection and Ranging (LiDAR), were used to create Digital Elevation Models (DEM). Finally, Ground Penetrating Radar (GPR) was useful to create a data cloud of possible subsurface damages. Machine learning technique (clustering) was applied to the GPR data to assist in defect depth analysis, further improving understanding and knowledge. The integration of complementary non-destructive methods offers a comprehensive framework for advancements in road infrastructure analysis and maintenance strategies, thus contributing to decision-makers to enhance transportation infrastructures' resilience and longevity.

This paper presents a methodology for a low – cost 3D visualization of property rights in multi – storey urban buildings using available and other relevant open geometric and legal data. An online, free platform is created, named “CadaSPACE”, to support the homogenous visualization of volumetric property units and open property rights included in 2D cadaster. The platform is directly interconnected with open – access pages such as the National Cadastral Portal, Autodesk Online Viewer, Tandem, Google Earth and Online 3D Viewer. It offers a wide range of statistical tools such as tables and diagrams to present, filter and manage the sematic information of the 3D property volumes. An application is presented for the Greek Cadaster. Legal information about the property rights is collected from the open - access cadastral records; building footprints are digitized on the available Orthophotos; architectural floor plans of each individual property unit are either taken from the documents included in the electronic building identity records (under construction) or are provided by the owners, and an approximate BIM (Building Information Model) is created for each building using additional approximate geometric information derived from Google Earth Pro and Streetview. The 3D property units are modelled and visualized as volumes. The whole 3D neighborhood model is uploaded online for further categorization and management through various visual and semantic filters. Considerations for further legal and technical improvements are given towards the implementation of the proposed methodology.

Figure-ground maps play a key role in many disciplines where urban planning or analysis is involved. In this context, the automatic generation of such maps with respect to certain requirements and constraints is an important task. This paper presents a first step towards a deep automatic generation of figure-ground maps where the built density of the generated scenes is controlled and taken into account. This is preformed building upon a Geographic Data Translation model which has been applied to generate less available geospatial features, e.g. building footprints, from more widely available geospatial data, e.g. street network data, using conditional Generative Adversarial Networks. A novel processing approach is introduced to incorporate the population density and the built density accordingly. Furthermore, the impact of both the level of detail of the street network, i.e. its sparsity or density, and the spatial resolution of the training data on the generated figure-ground maps has been investigated. The generated maps and the qualitative results reveal an obvious impact of these parameters on the layout of built and unbuilt areas. Our approach paves the way for the expansion of existing districts by figure-ground maps of future neighbourhoods considering factors such as density and further parameters which will be subject of future work.

Wayfinding in large indoor spaces is challenging for the limited views the wayfinder has. Verbal indoor wayfinding assistance converts the weakness into an advantage: This experimental study investigates whether such a conversation-based approach based on what is in view can lead to successful wayfinding in complex indoor environments. The findings of this experimental study indicate that by using verbal guidance people can arrive at their destination successfully without experiencing ambiguity of decisions and hesitation of decision-making time, nuanced by a set of environmental and personal attributes. These attributes are captured by variables including the route instruction grammar, familiarity with the environment, and configurational information.

Sponge cities are renowned for their efficacy against extreme weather events, reducing surface runoff, managing stormwater, and mitigating flood risks. Moreover, they present multifaceted advantages by integrating blue-green infrastructure, enhancing urban sustainability, and improving water quality. The trend of their expansion beyond China marks a significant development in climate-resilient urban planning. This study pioneers the use of open data to locate suitable sites for sponge Low Impact Development (LID) solutions, showcasing Vigo (Spain) as a viable case for mid-sized cities. Input data is obtained from administrative cartography (DTM, hydrogeology, land cover, river courses, and demographic census) and satellite imagery (impervious coverage, vegetation, and surface temperature) from Landsat 8 and MODIS calculating three spectral indices (NISI, NDVI, NDIH). A robust Geographical Information System (GIS) method is proposed weighting the multi-criteria with AHP matrix. Three main potential sites are identified for deploying specific sponge LID strategies, as green roofs, green parking, or rain gardens. Nevertheless, while the method swiftly identifies intervention sites on a municipal scale, conclusive decisions necessitate terrain insights, public sentiments, urban regulations, and funding considerations.

Numerous navigation applications rely on data from global navigation satellite systems (GNSS), even though their accuracy is compromised in urban areas, posing a significant challenge, particularly for precise autonomous car localization. Extensive research has focused on enhancing localization accuracy by integrating various sensor types to address this issue. This paper introduces a novel approach for car localization, leveraging image features that correspond with highly detailed semantic 3D building models. The core concept involves augmenting positioning accuracy by incorporating prior geometric and semantic knowledge into calculations. The work assesses outcomes using Level of Detail 2 (LoD2) and Level of Detail 3 (LoD3) models, analyzing whether facade-enriched models yield superior accuracy. This comprehensive analysis encompasses diverse methods, including off-the-shelf feature matching and deep learning, facilitating thorough discussion. Our experiments corroborate that LoD3 enables detecting up to 69% more features than using LoD2 models. We believe that this study will contribute to the research of enhancing positioning accuracy in GNSS-denied urban canyons. It also shows a practical application of under-explored LoD3 building models on map-based car positioning.

The development of a knowledge-based decision support system for the evaluation and planning of location and urban development concepts was implemented. In order to achieve this goal, cross-domain ontologies were developed for interdisciplinary databases, which are then mapped in semantic networks. The exponential growth in computing power in the hardware sector alone can no longer solve this problem, but at the same time enables the application of new methods for storing and evaluating data. Essentially, it is no longer just about the digital recording of object properties in conventional databases, but also about the digital representation of their significance for specific questions and the linking of meanings across the boundaries of specialist domains. This information is stored in a multimedia knowledge base, together with the methods and rules for its use and the evaluations and decisions based on it. The motivation for this project is the rapidly growing amount of data, which extends across ever new specialist domains and can no longer be sufficiently integrated into the decision-making of experts using conventional methods of knowledge acquisition. After determining this data, it was linked to a georeferencing. Within the framework of the project, documents were analyzed with the help of AI and examined for semantic text corpora. This data was georeferenced. Various algorithms were used to accomplish this task, including TF-IDF, TextRank and Word2Vec.

3D point clouds acquired with terrestrial or mobile LiDAR sensors are increasingly used to map urban forests. The segmentation of separate tree instances, i.e., subsets of points representing individual trees, is a relevant step in automatically extracting tree inventory data from 3D point clouds. Various algorithms have been proposed for tree instance segmentation, offering different trade-offs between accuracy, runtime, and robustness against data incompleteness and noise. In this work, we propose a coarse-to-fine algorithm for segmenting tree instances in urban 3D point clouds from terrestrial or mobile LiDAR scanning that combines two existing techniques: (1) the computationally efficient marker-controlled Watershed algorithm and (2) a more accurate 3D region growing algorithm. Initially, the marker-controlled Watershed algorithm generates a coarse segmentation, which is further improved by a Voronoi segmentation-based error removal. Subsequently, the coarse segmentation is refined by the 3D region growing algorithm in areas with overlapping tree crowns and sufficient data quality. In both steps, our algorithm uses the results of a prior semantic segmentation to select suitable markers and seed points. We evaluated our coarse-to-fine algorithm in an ablation study using two mobile LiDAR datasets and one terrestrial LiDAR dataset from three German cities. Our results show that our algorithm outperforms the standard marker-controlled Watershed algorithm in terms of panoptic quality by 3.7, 25.5, and 29.6 percentage points, respectively, while being computationally more efficient than an approach purely based on 3D region growing.

The incorporation of detailed textures in 3D city models is crucial for enhancing their realism, as it adds depth and authenticity to the visual representation, thereby closely mimicking the surfaces and materials found in actual urban environments. Existing 3D city models can be enriched with energy-related roof and façade details, such as the material type (such as windows, green façades, bricks) and sunlight reflectance which can be derived from texture information. However, a common limitation of these models is their lack of very high resolution textures, which which reduces their realism and detail. Manually mapping textures onto each surface of a building is an exceptionally time-consuming and labor-intensive process, making it unfeasible for large-scale applications involving thousands of buildings. Therefore, an automated method is essential for texture mapping of 3D city models from aerial imagery. In this paper, we present CityJSON texture mapper – a python-based software tool for automated texture mapping of CityJSON-based 3D city models from oblique and nadir aerial imagery. Experimental results demonstrate the effectiveness of our approach in generating high-quality textured 3D city models, showcasing the potential for broader applications in geospatial analysis and decision-making. This research contributes to the ongoing efforts in enhancing the realism and usability of CityJSON-based 3D city models by enhancing them with their real textures from oblique aerial imagery. Texture mapped model can be explored at https://bit.ly/textured3dbag.

This paper addresses the challenge of managing and maintaining road pavements effectively throughout their lifecycle. Emphasizing the need for a cohesive data collection and handling approach, the paper highlights the vulnerability caused by the absence of a structured information system, resulting in reworks, information loss, and misinterpretation of collected data. A unified data handling and collection structure, introducing the concept of a pavement Digital Twin through a standardized data structure is presented. The Digital Twin aims to integrate information regarding pavement failures, with the future aim of predicting deterioration and facilitating informed decision-making in management and maintenance interventions. Complexities in pavement failures, categorized into surface and subsurface modes, prompt the necessity for a reliable classification and representation system. The proposed methodology introduces a grid of cells for surface failures and three-dimensional voxels for subsurface failures, providing a structured approach for data integration throughout the pavement's lifecycle. An approach for the semantic representation of pavement distresses is also presented. The proposed methodology stresses the importance of scalability and flexibility in data storage, forming the basis for a comprehensive Digital Twin of road pavements. Finally, through the use of multiple pavement failure datasets, the methodology is shown to have a high potential in providing a structure for comprehensive management of road data.

Autonomous driving and traffic flow simulation requires a realistic and accurate representation of the environment. Therefore, this research focuses on the semantic segmentation of aerial images for simulation purposes. Initially, a dataset was created based on true orthophotos from 2019 and Kempten’s street cadaster, with true orthophotos being fully rectified aerial images. The chosen classes were oriented towards the subsequent conversion and usage in simulation. The proposed labeling workflow used cadaster data and demonstrated significant time efficiency compared to state-of-the-art datasets. Subsequently, a neural network was implemented that was trained and tested on the dataset. In addition, the network was also trained only on the lane markings to compare the network’s performance. Both cases demonstrated excellent segmentation results. The generalizability was then tested on true orthophotos from 2021. The results indicated a solid generalizability, but still needs to be improved. Finally, the aerial information was converted into a 3D environment, that can be used in simulations. Our results confirm the usage of aerial imagery and street cadaster data as a basis for the simulations.

This work presents a preliminary case study on assessing reinforced concrete pedestrian bridge conditions in A Coruña (Galicia, Spain), employing a novel integration of non-destructive testing (NDT) technologies. The research aims to refine the health monitoring process of pedestrian bridges by adopting a top-down approach, leveraging the data-fusion concept to enhance the analysis of structural defects. A data-fusion methodology that integrates interferometric synthetic aperture radar (InSAR), ground penetrating radar (GPR), and light detection and range (LiDAR) technologies is introduced. Then, the data are visualized within a 3D Geographical Information System (GIS) environment. The MT-InSAR technique is used at the network level to identify bridges requiring detailed inspections. Subsequently, selected bridges undergo thorough examinations using GPR and LiDAR technologies from fieldwork between 2021 and 2023. A comparative analysis of three different LiDAR devices and two GPR setups is conducted to evaluate their effectiveness in capturing detailed structural data. The study also explores the integration challenges and solutions for combining diverse data formats and the possibility of using advanced digital technologies, such as Building Information Modelling (BIM), to facilitate a seamless transition from traditional NDT approaches to a digitized, model-based inspection framework. The findings highlight the advantages of each NDT method, address specific data acquisition challenges, and propose strategies for overcoming issues related to data integration, visualization, and the accuracy of spatial localization. Integrating these NDT technologies within a georeferenced 3D GIS environment facilitates a detailed understanding of the bridge's condition and enhances decision-making processes for maintenance and rehabilitation efforts.

The development of novel and robust digital methods to support the maintenance of existing road infrastructure requires a large amount of harmonized data. Especially in the context of automated modelling having a large amount of matching data from different perspectives enables disruptive, new use cases that might largely impact the efficiency in maintenance of the built environment. Unfortunately, such data compositions are tedious to collect in real world applications, due to many influential factors, leading to deviations between multiple data sources and the sheer complexity in the process of creating a digital model. However, for deep learning applications, a large amount of carefully annotated data is necessary for robust estimations. In this contribution, we tackle this problem by presenting a novel procedural modelling and model configuration approach for generating homogeneous data combinations to step towards direct parameter estimation for machine learning approaches utilizing point clouds of roads and end-to-end model generation of digital road models.

Uncrewed aircraft systems (UAS), also known as drones, have become increasingly popular in various applications due to their ability to access remote or challenging locations. A crucial aspect of UAS operations is flight path planning, which determines the trajectory the aircraft takes to achieve its mission objectives. However, the diverse use cases of UAS demand different planning approaches tailored to their specific requirements. This paper presents an overview over current research in UAS flight path planning and proposes a categorization of use cases based on their distinct goals and considerations. The terminology differentiates the goals of tasks between navigation-centric and data acquisition and also considers their context, domain, perspective, and sensors. We analyze existing approaches with a focus on UAS flight path planning for data acquisition using cameras, highlighting their strengths and limitations. This structured overview facilitates the understanding of the diverse landscape of UAS flight path planning and paves the way for the development of more targeted and effective solutions for various applications and their use cases. The analysis shows, that the term ”UAS flight path planning” is currently used in a variety of distinct use cases with diverging requirements, so a unified terminology should be established for clear communication.