MSc Thesis topics

This list gives an overview of potential graduation topics for MSc Geomatics and GIMA students.

Filter

Show those topics, that match any of the following:

By keyword
By contact
Table of Contents

3D mesh generation from Oblique Photos with AI (at Geodelta) 🔗

Description

This research project (graduation internship at Geodelta) focuses on developing advanced methodologies for generating 3D meshes from oblique photographs using artificial intelligence. Leveraging the 'Segment Anything' deep learning network, the project aims to accurately segment buildings and other identifiable features within oblique images.

Objectives and challenges for this project include:

  1. AI-Based Image Segmentation: Utilizing the 'Segment Anything' deep learning network to identify and segment buildings and other key features in oblique photographs with varying perspectives and lighting conditions.
  2. 3D Mesh Post-Processing: Combining segmented features from multiple images, using their known positions and forward intersections, to reconstruct detailed and accurate 3D meshes.
  3. Urban Visualization Enhancement: Developing techniques to enhance 3D visualizations of urban environments, aiding municipalities in city planning and management.

3D mesh generation

Geodelta is looking for a motivated student who is willing to make a deep-dive in photogrammetry and software engineering. While we are happy to provide a lot of support, the following competences will give you a head-start in your internship:

At Geodelta, we know that measuring is more than just precision. It is the reliability that counts. Are the measurements correct and have the correct procedures been applied? We are engineering consultants focussed on delivering the best technical advice and software in the fields of geodesy, photogrammetry and laser scanning. We make sure that our clients can trust their measurements. We do this by developing user-friendly software and by providing trusted advice. We are independent, innovative, pragmatic and above all we have a passion for geodesy. We are based in the century-old former faculty building for Geodesy. This is just around the corner of the building of the faculty of Architecture. If you are interested in this topic, or if you have an amazing idea yourself, let us know! We will arrange meetings with interested students discuss the project(s) and find the best fit.

References

A. Kirillov et al., "Segment Anything," 2023 IEEE/CVF International Conference on Computer Vision (ICCV), Paris, France, 2023, pp. 3992-4003, doi: 10.1109/ICCV51070.2023.00371.

Contacts

Keywords

Artificial Intelligence, Digital Twins, Imagery

3D zoning plans 🔗

Description

Covering existing zoning plans into 3D (similar to the Indonesian case of Jakarta and Bandung) and automatically including this in design constraint and checking building permit requests. Based on the obtained experience also guidelines for the direct 3D zoning plans in the future should be developed.

Contacts

Keywords

Land Administration

A digital twin based on Land Administration 🔗

Description

A digital twin can be defined as a digital representation of a physical asset, its related processes, systems and information, including physical assets indoors, outdoors, above and below ground. Digital twins combine state-of-the-art engineering models and analytics with asset specific operational data to create digital simulation models and information models that are updated and changed throughout the lifecycle of their physical counterparts. A Digital Twin provides a scope for efficient information management and multidisciplinary collaborations for different decision-making objectives in a built environment.

The digital twin is a concept that will enhance information management and collaboration, where experts can work together, preventing costly mistakes and rework. It is important to state that a Digital Twin also encompasses 4D, the time; thus, maintaining date and time attributes.

An MSc thesis topic could focus on the creation of the digital twin of a use case area (i.e. 3D models in differentLoDs ), including key design and condition information for physical assets above and below ground, plus legal boundaries to better manage current and future developments. Complex digital information about the area's infrastructure will be modelled according to international standards (ISO, OGC, industry standards, etc.) and visualised using 3D web-viewers (e.g. Cesium, etc.). One of the objectives could be to outline strategies and methodologies to link the virtual and legal (real) Rights, Restrictions and Responsibilities (RRRs), land use, etc.

3D_Cad4.jpg

Related literature :

Contacts

Keywords

Digital Twins

A model-based approach for layout reconfiguration from indoor point clouds 🔗

Description

Source: https://www.pinterest.com/pin/233905774372839617/

Layout problems range in scale from urban planning by the assignment of activities to cities, sites, or buildings, to the location of equipment and personnel on a single floor, or even to the location of components in printed circuit boards. Reconfiguring existing indoor spaces is also typically addressed in interior design and building renovation.

This MSc thesis will focus on the development of a method for the automatic layout reconfiguration of indoor scenes from point clouds. For this purpose, model-driven approaches will be explored to introduce rules aiming to automatically reconfigure the position of building objects according to their function. Point clouds captured with Apple Smart Devices will be explored as data source.

Required skills: Proficient in Python programming

Contacts

Keywords

Point clouds

Adjustment of 20 million boundaries 🔗

Description

For this subject we are looking for a master student with special interest in geodesy and surveying to help to solve the challenge of a complete adjustment of all the boundaries in the Netherlands in order to improve the geometric quality of the cadastral map to a sufficient level for reconstruction (showing the invisible boundary). That means that the gigantic amount of information stored in 20 million pdf-documents must be transformed to input data for an adjustment program and be compared with other large scale topographic and photographic. In particular your interest will be in designing an adjustment process, finding a practical way of describing the geometric quality and finding a way of automatic quality verification.

Contacts

Keywords

Land Administration

ahn-ahn2-ahn3-ahn4-ahn5 temporal-delta analysis 🔗

Description

Contacts

Keywords

Point clouds

Air Rights Restrictions and Responsibilities: investigate, model, and visualise them in the context of ISO 19152 LADM 🔗

Description

The fact that a 3D property unit must comprise a building or other facility or a part of it, does not mean that the property unit has to be developed when it is formed. During a transition period, property units that consist of airspace or unused space below ground can be permitted, bought, sold, etc. Air-space parcels are 3D parcels created vertically and horizontally within a building to establish separate legally transferable lots. This means that there is need to investigate the types and modelling approaches of air Rights, Restrictions and Responsibilities attached to those spaces as well as their visualisation, as they also contribute much of the value of the spatial unit. This procedure is subject to the legislative framework and regulations of each country and limited by local building codes, ordinances, and zoning restrictions. The last years, the spatial management of complex urban environment is asking for 3D registration and representation of Rights, Restrictions and Responsibilities (RRRs). The first edition of the ISO 19152 Land Administration Domain Model (LADM), one of the first spatial domain standards, providing a flexible conceptual schema as basis for the development of 2D and 3D cadastres based on a Model Driven Architecture (MDA) (ISO 19152, 2012), allows for all those levels of complexity (2D, 3D, as well as integrated 2D and 3D spatial units) with various levels of accuracy. Currently, the standard is under revision, and it is a good opportunity to initiate the discussion on the registration of air-parcels. Objective: This MSc thesis could focus on the modelling of 3D air-space parcels, in the context of LADM revision and explore visualisation options.

3D_Cad5.jpg

Related literature :

Contacts

Keywords

Land Administration

Allmaps - Curating, georeferencing and exploring digitized maps 🔗

Description

Allmaps logo

Introduction: Allmaps is an innovative open-source ecosystem dedicated to curating, georeferencing, and exploring digitized maps. It leverages the standards of the International Image Interoperability Framework (IIIF) used by a large number of cultural heritage institutions worldwide to present their digital collections. By offering accessible web-based interfaces, Allmaps introduces GIS to new audiences while retaining compatibility with traditional methods. Together with an international consortium of institutions, TU Delft Library supports Allmaps’ development with the aim to make digital collections more usable for research and education.

Potential thesis topics can contribute to the project by:

NB: Allmaps consists of a series of Javascript packages, Svelte applications, a PostGIS database and a Command Line Interface. While projects making use of Python or other programming languages are not discouraged, they might benefit less from the available modules and cannot directly contribute to the codebase in the form of pull requests.

Below is a list of concrete suggestions for thesis topics; feel free however to get in touch to suggest your own.

Thesis Topic: An Interactive Trade Directory of 19th Century Paris

The first part of the research will focus on georeferencing digitized cadastral maps of Paris from the early 19th century held by the Bibliothèque nationale de France (BnF). For this, image segmentation and existing shapefiles of each building block will be used. The second part involves relating the georeferenced maps to data extracted and enriched from 19th-century trade directories by the SODUCO project. This enables the identification and analysis of specific businesses and their corresponding building plans.

Links:

Single cadastral sheet opened in Allmaps Viewer

Thesis Topic: Modeling and Metadata Description of Map Series

This topic will focus on developing a shared metadata model for historical map series, taking into account different (sub)series, editions and individual sheets, their logic and exceptions. The aim is to standardize this information using Linked Open Data principles. The second part involves applying the developed model to enrich the metadata of existing digital editions, spread over multiple institutions, to demonstrate its effectiveness and applicability.

Links:

Overlaid sheet indices of Dutch map series in Bonne projection

Thesis Topic: Exploring the Challenges and Solutions of Georeferencing Oblique Aerial Views and Architectural Elevations and Sections.

This topic will delve into the process of 3D georeferencing, focusing on oblique aerial views (photographs and drawings) and architectural drawings, such as elevations and cross sections. The research will identify and analyze the problems encountered, such as those related to scale, location, the data model, and assessing the accuracy or uncertainty of results. Furthermore, it will explore the interfaces required for effective georeferencing, viewing, and interacting with these resources. The aim is to provide a comprehensive understanding of the complexities involved and to propose innovative solutions to overcome these challenges.

Links:

Los Angeles 1909 (Los Angeles Public Library WALLMAP-00102)

Thesis Topic: Automated Georeferencing of Topographical Map Series: A Machine Learning Approach

This topic will focus on georeferencing various map series held by the Dutch National Archives using machine learning and feature matching techniques. The research will consider different characteristics of the maps, ranging between homogenous layouts and scattered loose sheets. It will also explore the potential of extending the approach to other map series such as those made of the former Dutch colonies. The project will continue and enhance the work done by previous students (Geomatics Synthesis Project) with the ultimate goal to complete the georeferencing of the 4.TOPO archive.

Links:

Plot of feature matching from Geomatics Synthesis Project report

Contacts

Keywords

3D, Artificial Intelligence, Georeferencing, Geoweb, Historical maps, Imagery, Standardisation

An xJSON format for ISO 19152 LADM implementation 🔗

Description

As an international standard, the ISO 19152 LADM may stimulate and accelerate the implementation of land administration systems. The LADM is already utilized in various applications and implementations worldwide, usually through the elaboration to a country profile (e.g. UML application schema), and the realization of this profile with a physical (technical) model suitable for implementation: a database schema (Data Definition Language --DDL-- schema of an intended database), a data exchange format (GML,GeoJSON , RDF) and a user interface for edit and dissemination. The LADM is currently under revision and, in addition to the extension of the various parts of the conceptual model defined in Edition I, attention is paid on the steps towards implementations. In addition to the conceptual model, the intention is that Part 6 - LADM Edition II will also include the corresponding technical models (CityGML, InfraGML, RDF, INTERLIS, BIM/IFC, GeoJSON). In this regard, and as JSON formats are gaining ground, it is expected that the multipart Edition II of LADM can be directly implemented in xJSON files to be used in webbased applications. Objective: The aim of this MSc this is the development of an xJSON format as an alternative technical encoding for the implementation of ISO19152 LADM supporting both spatial and non-spatial information.

References:

Contacts

Keywords

Land Administration

Assessment of the effect of outdoor climate on cities' heating energy 🔗

Description

Buildings are responsible for about 40% of the national energy use . They are therefore a main target of energy transition policies. The availability of public energy data at Zipcode level 6 all over the Netherlands makes it possible to map geographically the energy use (see for instance http://www.energieinbeeld.nl/). There are large differences between cities and neighborhoods that can partly be explained by socio-economic factors and construction types. Another factor that has not been research extensively yet is the influence of the climate.

Energy models usually use climate data from KNMI meteorological stations. Generally, the station De Bilt is used as reference climate, but other stations can be used as well, and even networks of individual home stations. This thesis should answer the following questions:

  1. What are the difference between diverse meteorological stations in terms of temperature, wind and humidity?
  2. To what extend can this differences explain differences in heating energy use at neighborhood/city level?
  3. Is it possible to use locale weather data sources (e.g. networks of home stations) en does then the explanatory power of weather to explain variations in heating energy use increase?
  4. Is it possible to produces maps with heating energy use corrected for weather influence?
  5. Is the usually used Degree Day correction method in agreement with the findings in this research . If not can we propose a better method?

Contacts

Keywords

Digital Twins, Energy

Building a new "terrestric registration" 🔗

Description

Building a new registration in which the surveydata and the map will be fully integrated is a very import step in the described process. The challenge is to find a datamodel that serves all the functionalities required for this goal. Mapping boundaries and parcels may not look too complicated at first sight, from a closer look however there are many exceptions that have to be taken in account. Because of the legal aspects of the registration many details must be stored, often also historical details.

Furthermore we need to take into account that by rebuilding the cadastral map with the accurate survey data we need to store the information in a way that automated mapping is possible. This means that the definitions of the entities must be very adequate.

A third requirement is the development of a three dimensional cadastre in the future. The datamodel which have to be designed must take this into account and reserve enough space for attributes.

For inspiration the international (LM-KAD) and national (IM-geo) standards will be used. We need a masterstudent to help us designing a suitable datamodel and to give advice of how to build it. Together we will define the exact scope of the thesis and make sure that you will get a distinguished part of the design one can attribute to you.

Contacts

Keywords

Land Administration

Building a tool for automated calculation and mapping 🔗

Description

The project cadastral map improvement does risk becoming much too expensive when one has to perform the conversion of the data by hand. For that reason we intend to develop a tool which can automatically select the information from a database and transform that data into input for the geodetic calculation of the new boundary coordinates. The greatest challenge is to find a tool which can automatically interpret handwritten field sketches. The artificial intelligence needed to reach this goal is obtained by adding all kinds of information together in order to identify the different positioning points in the process.

The scope of the master student will be in designing GIS-solutions in the described process. There will be many connections to the different tasks in this process: to the interpretation software, the geodetic adjustment software and to the database that will be designed in the process by subject 1 (terrestric registration). The data itself has many variety in quality, depending on the moment in history boundaries were established.

You will be part of the designing team and work together with an intern who already finished a bachelor program in surveying and geomatics. She will focus on testing this process both manually and with prototypes. Of course, here also we will define the exact scope of the thesis together and make sure that you will get a distinguished part of the design one can attribute to you.

Contacts

Keywords

Land Administration

Bundle-Block Adjustment with Panorama Imagery (at Geodelta) 🔗

Description

This research project (graduation internship at Geodelta) aims to advance the precision and accuracy of geospatial measurements using photogrammetry and bundle-block adjustment techniques applied specifically to panoramic street-view images. Many municipalities in The Netherlands have extensive collections of street-view panorama pictures; however, the current positioning of these images is not sufficiently precise for accurate measurements.

Objectives and challenges for this project include:

  1. Feature Detection in Panorama Imagery: Developing methods to accurately detect and match features within panoramic images.
  2. Collinearity Function Adaptations: Modifying the existing collinearity equations used in bundle-block adjustment to accommodate the unique characteristics of panoramic imagery.
  3. Precision and Accuracy Evaluation: Finally and maybe most importantly, assessing the impact of the implemented procedures on the precision and accuracy of the resulting geospatial measurements.

This project will provide valuable insights and methodologies that can significantly improve the usability of panoramic street-view images for precise geospatial measurements, benefiting urban planning and municipal management.

Bundle Block Adjustment

Geodelta is looking for a motivated student who is willing to make a deep-dive in photogrammetry and software engineering. While we are happy to provide a lot of support, the following competences will give you a head-start in your internship:

At Geodelta, we know that measuring is more than just precision. It is the reliability that counts. Are the measurements correct and have the correct procedures been applied? We are engineering consultants focussed on delivering the best technical advice and software in the fields of geodesy, photogrammetry and laser scanning. We make sure that our clients can trust their measurements. We do this by developing user-friendly software and by providing trusted advice. We are independent, innovative, pragmatic and above all we have a passion for geodesy. We are based in the century-old former faculty building for Geodesy. This is just around the corner of the building of the faculty of Architecture. If you are interested in this topic, or if you have an amazing idea yourself, let us know! We will arrange meetings with interested students discuss the project(s) and find the best fit.

Contacts

Keywords

Imagery

Cadaster and Madaster 🔗

Description

Background and aim: The built environment is increasingly becoming spatially complex, where shaping and sharing AECOO, spatial and economic data into an efficient data flow represent a challenge. The potential for the reuse of information within the Spatial Development Lifecycle (SDL) is a significant factor in calculating its economic value. Therefore, in the scene of SDL, digital documentation of materials for real estate objects is very important, taking into consideration EC Digital Building Logbooks approach, as well as the international standard ISO 19152:2012 Land Administration Domain Model (LADM). Information stored in materials passports is only useful when it can be used by the relevant actors at the required time, and therefore, need to be integrated into BIM to provide input data for assessments on reversible and circular design .

Research questions: How can materials passport be reused for land administration applications?

madaster.png

Research objective: This research aims to investigate the potential relationship between Cadastre (or 3D Land Administration System in broader sense) and materials passport (Madaster/ cadastre for materials). To reach this aim, the following steps need to be undertaken: 1) Define specific application areas where Cadaster and Madaster synergy could be beneficiary within the SDL. 2) Investigate how Life Cycle Assessment (LCA) Databases can be linked with Land Administration Systems (LAS). 3) Provide an outlook on challenges to be tackled in the future considering the BIM availability and circular economy.

Method: Literature research; Investigate potential application areas; develop and validate creative design solutions; elaboration of a case study.

References:

Contacts

Keywords

3D, Land Administration

Cadastral map quality 🔗

Description

[Description from Dutch cadastre] Cadastral information is a crucial precondition for the economic development of a country. Since 1832 when the Dutch Cadastre was founded, nearly every boundary is described by a so called "field sketch" in which detailed measurement-info is stored in order to be able to show the boundary to an owner at any time. Every parcel is also described in the so called "cadastral map".

This cadastral map does have a sufficient quality for the purpose it is build for: a viewer that serves as an entry into the cadastral registration. The map is complete and not divided into map sheets anymore. The geometric quality of the map is guaranteed with a point precision of about 20 cm for builded areas and 40 cm for rural situations. More and more we have problems with this quality: people does expect a quality like we have stored in the field sketches which are specially created by surveyors to be able to point out a very accurate position of a boundary.

Therefore the Dutch Cadastre decided to transform the high geometric quality of the field sketches into the map and make it more accurate. A year ago this innovation research was started. Within this project we have defined some subjects who need a more detailed research

veldwerk.png

Contacts

Keywords

Land Administration

Change detection using LoD 🔗

Description

Context: High accuracy and easy collection make point clouds an ideal data source for change detection. However, the large data size causes problems for efficient comparison between point clouds, e.g., AHN2 and AHN3. Level of Detail (LoD ), as you have learnt, is an effective technique to address the big data issue. We choose certainLoD depending on the accuracy requirement of the problem: once more details are needed, we employ a lowerLoD . The general rule is of course, using as less points as possible, to achieve high efficiency. However, how the [ LoD influences the accuracy of change detection is unknown. Thus, this project aims at investigating the possibility to adopt aLoD approach to improve change detection on large point clouds.

imgChange1.jpg imgChange2.jpg imgChange3.jpg

Objective: Main research question: What can we gain by using LoD for change detection on large point clouds? Sub-research questions 1) What types of changes should be concerned? 2) What features should an optimalLoD have, for change detection? For example, it should be simple to build/compute, yet it can detect as many changes as possible when using same number of points. 3) How doesLoD influence the results of change detection? Proper metrics for evaluation should be proposed first. Technical route is depicted as follows,

imgChange4.png

You will start from the Octree structure, a widely usedLoD structure. First go through the whole process. After understanding the problems, try to improve theLoD strategy adopted. In the end, sound comparison will be conducted and presented to indicate the benefits from theLoD approach. For this project, you do not have to start from scratch, as many open source codes are available.

Contacts

Keywords

Change detection, Point clouds

Compare solutions for bringing 3D subsurface data on the web (at Geological Survey of the Netherlands) 🔗

Description

Subsurface illustration

Geological Survey of the Netherlands (GDN), part of TNO, has a total of 285 employees working in four departments. As a knowledge partner to the Dutch government, GDN has three main tasks: we collect, manage and provide access to information on the subsurface (information task), we advise governments on the use of the subsurface (consultancy task), and we develop applied knowledge on the subsurface and subsurface-related technology (research task). We employ geologists, physical geographers, engineers and, because of our expertise in data management, relatively many geodata scientists and ICT specialists.

As part of our statutory duties under the Mining Act and the Subsurface Base Registration Act (BRO), GDN collects all kinds of (research) data about the subsurface. We manage these data and make them available via portals such as the BROloket, the DINOloket and NLOG. We also create and disclose digital models -GeoTOP , REGIS and DGM- of the subsurface for various applications, which we disseminate through the same channels.

The demand for information from our users - the government and engineering firms - is increasing and particularly for an easily accessible access to 3D models. Soon models in the BRO beingGeoTOP , REGIS and point data such as cone penetration tests and groundwater wells with positions will be available nationwide as ESRI Indexed 3D Scene Layer (I3S ). We realize that there are more developments in the field of 3D online visualizations. With this research we want to compare different standards/implementations. Think aboutCityGML /JSON, Cesium 3D tiles and applying them in for example the Unity platform.

Note: because the documentation of the underground information is mainly available in Dutch, a good command of the Dutch language is required.

Contacts

Keywords

Geoweb

Comparing compression techniques for 2.5 and 3D point clouds 🔗

Description

DracoLogo.jpg

A number of specialized compression techniques exist for storing point clouds: Laszip, Google draco, Esri lerpcc General compression codecs are also available, e.g. when storing point clouds in Apache Parquet format, like Brotli, zstd, snappy, gzip, lzo, brotli, lz4_raw.

Questions that should be central in this research project:

References

Contacts

Keywords

Point clouds

Converting segmented point clouds into enriched navigable maps for pedestrians in urban scenarios 🔗

Description

image_segmented_pc.png

Figure: a) Semantically segmented point clouds; b) an urban map extracted from OSM; c) different spatial granularities of maps extracted from semantically segmented point clouds. In summary, in the last years much effort was made to semantically segment point clouds. However, most of mobility tools for pedestrians are still based on networks created from road centrelines instead of representing sidewalks and pedestrian crossings. Consequently, there is still a need of representing the navigable space for pedestrians in a more accurate way to enable realistic path finding. In the recent years, a few of papers proposed the use of point clouds for updating the navigable network. Some papers propose the direct use of classified point clouds for path finding but this solution is impracticable for urban scenarios. This MSc topic propose the conversion of segmented point clouds into enriched networks enabling path finding for pedestrians in urban scenarios. Students will work on segmenting the navigable space for pedestrians, converting the point clouds into networks with different spatial granularities and attributes, and integrating the result into OSM.

Requirements and skills: programming

Contacts

Keywords

Point clouds

Cyclomedia data benchmarking 🔗

Description

Contacts

Keywords

Point clouds

Data fusion point clouds: nat+droog+kunstwerk 🔗

Description

Contacts

Keywords

Point clouds

Design and improvement of self-adaptive 3D structuring elements for mathematical morphology in point clouds 🔗

Description

Context: Many image processing techniques have been adapted in recent years for use in point clouds: region growing, connected components, Convolutional Neural Networks, etc. The mathematical morphology has also been adapted for use in point clouds, although, comparable to image processing, the structuring element in a point cloud is strongly related to orientation and point density. The use of structuring elements with fixed orientations and point densities limits the applicability of mathematical morphology to point clouds with unknown orientations and strong density changes.

bunny.png

Objective: This work aims to design structuring elements that adjust in orientation and local point density to the point cloud in the processes of morphological dilation and erosion. Its application will be tested on real and artificial point clouds to solve problems of detection, segmentation and gap filling. During the whole process it will be very relevant the optimization and parallelization, given the great amount of iterations.

Contacts

Keywords

Point clouds

Designing a Data Infrastructure for Digital Twin Technology in Energy Management 🔗

Description

Background and Aim

The advancement of Digital Twin technology has significantly broadened the prospects for energy management, particularly within the realm of building operations and energy transition initiatives. A key to unlocking these opportunities is the establishment of a robust data infrastructure capable of managing complex energy-related datasets efficiently. This research project aims to design and implement an advanced data infrastructure that supports an integrated energy model, facilitating enhanced energy management and decision-making processes.

Research Questions

  1. What are the key components of a data infrastructure that support effective energy management using Digital Twin technology?

  2. How can data and information modeling be optimized to improve energy management in buildings?

  3. What role do standardization and integration play in enhancing the functionality and interoperability of digital twins for energy management?

  4. How can semantic technologies, including ontologies and knowledge graphs, contribute to more sophisticated data discrimination and integration?

Research Objectives

This research intends to create a sophisticated framework for data infrastructure tailored to energy management through Digital Twin technology, with the following specific objectives:

  1. To develop a comprehensive data model that integrates diverse energy datasets, facilitating effective data management and analysis.

  2. To implement standardization practices that ensure data integrity and interoperability across different systems and platforms.

  3. To explore and incorporate semantic technologies to enhance data discrimination, ensuring accurate information extraction and usage.

  4. To construct an ontology-based knowledge graph that supports dynamic data linking and complex query capabilities, improving the insight generation from energy data.

Method

This research will employ a multi-disciplinary approach encompassing a thorough literature review, advanced data modeling, standardization practices, and the implementation of semantic technologies. Initially, a comprehensive review of existing literature on digital twins, data infrastructures, and energy management systems will establish a theoretical foundation. Subsequently, robust data models capable of managing complex energy-related datasets will be developed. These models will be standardized to ensure interoperability across various systems and platforms. Additionally, semantic technologies will be explored and utilized to design ontologies and knowledge graphs, which will support advanced data discrimination and integration. This integrated approach will culminate in the development and testing of a prototype to assess the practical effectiveness and scalability of the proposed data infrastructure in real-world energy management scenarios.

References

Contacts

Keywords

Digital Twins, Energy, Knowledge Graphs

Detecting building element/material through ground-based thermal imagery using Deep Neural Networks approach 🔗

Description

Identifying and analyzing the elements/materials used in construction supports promoting sustainable practices in the building industry. This process aims to enhance the lifecycle of materials by facilitating their reuse, recycling, and repurposing, thereby minimizing waste and reducing the environmental impact. Utilizing advanced technologies, both sensing technologies and Deep Neural Networks, provides the scope for more accurate and detailed detection of building elements and analysis of building materials. While mainly optical imagery and laser scanning data have been applied in detection of building elements (e.g. windows), the leverage of thermal imagery requires further research. A potential thesis topic is the leverage of close-ranged (ground-based) thermal imagery (using an existing thermal camera) for the detection of building elements (e.g. windows) and materials. Thermal imagery can be acquired with high temporal frequency for (part of) an existing building (e.g. Architecture Faculty building). A Deep Neural Network approach (e.g. (time-dependent) Convolutional Neural Networks) can be applied to the acquired thermal imagery for spatio-temporal analysis to enable the detection of building elements/materials.

Required Skill: Programming

Source: https://carelabz.com/building-envelope-infrared-thermography-inspection-service/

Contacts

Keywords

Artificial Intelligence, Imagery

Enhancing data reusability in 3D Land Administration Systems in line with ISO 19152 LADM II 🔗

Description

This thesis investigates methodologies to improve the reusability of data in 3D land administration systems (LAS), focusing on the use of LADM as a foundational standard. By leveraging the standards that can facilitate data reuse for land administration purposes, the research aims to develop and validate processes that enhance data interoperability and reusability within land administration.

The current limitations and challenges in the reusability of 3D cadastral data within existing LAS will be identified, focusing on technical challenges, while the refined survey model of ISO19152-2 will be used as basis and the results from the project "3D Cadastral Survey Data Model (3D CSDM)" will be considered. The potential impacts of improved data reusability on the efficiency and accuracy of 3D LAS will be explored, using case studies.

Related references

Contacts

Keywords

Database Management Systems, Land Administration, Standardisation

Enhancing Energy Demand Modeling with Spatial Data: A Geographic Approach to Simulation Accuracy 🔗

Description

Background and Aim: Accurate energy demand modeling is essential for effective energy management and planning, particularly as cities aim to increase energy efficiency and integrate more renewable sources. Current energy demand models primarily utilize data related to meteorological conditions, building specifications, and occupant behaviors. However, these models often overlook the influence of spatial factors --- such as the location of buildings relative to each other, urban density, and the proximity to infrastructural amenities --- which can significantly affect energy usage patterns. This research will focus on integrating and analyzing spatial data within energy demand simulations to improve their predictive accuracy and operational relevance.

Research Questions:

  1. How is spatial data currently utilized in energy demand modeling, and what gaps exist in its integration?

  2. What specific types of spatial data could enhance the predictive accuracy of energy demand models?

  3. What methodologies can be developed to incorporate spatial analytics into energy simulation models effectively?

Research Objectives:

The study will:

  1. Assess the current integration of spatial data in energy demand modeling, identifying existing limitations and opportunities.

  2. Propose a framework for systematically incorporating relevant spatial data (e.g., urban layout, building proximity, infrastructural characteristics) into energy demand simulations.

  3. Validate the proposed framework through empirical testing in varied urban environments, measuring improvements in model accuracy and reliability.

Method:

For the methodology, this research will adopt a dual approach to energy demand modeling by integrating both data-driven techniques and EnergyPlus models, enhanced through the incorporation of spatial data. he study will start with a comprehensive review of existing energy demand models to identify how spatial variables are currently used and to pinpoint gaps in data integration. Following the review, a hybrid framework will be developed that employs Geographic Information Systems (GIS) for spatial analysis alongside statistical and machine learning methods for data-driven modeling. This framework will be applied to both historical data and simulations conducted using EnergyPlus, a dynamic building energy simulation tool. The aim is to validate the improved accuracy of energy demand predictions through empirical testing across different urban settings, ensuring that both traditional and innovative modeling approaches benefit from the enhanced input of spatial data. This integrated methodology will allow for a robust comparison of model outputs, ultimately refining predictive capabilities and supporting more sustainable urban energy planning.

Contacts

Keywords

Energy, Simulation

Exploring BIM models for synthetic data generation towards indoor point cloud classification 🔗

Description

Recent integration ofLiDAR into smartphones opens up a whole new world of possibilities for 3D indoor mapping. These new systems offer an unprecedent opportunity for the democratization in the use of scanning technology. In fact, the large amount of available [ LiDAR data is arising interest by point cloud classification using artificial intelligence. Nevertheless, point cloud labelling is a time-consuming task and the amount of labelled data is scarce yet. In this context, data synthesis is gaining attention as an alternative to increase the volume of classified data.

Taking opportunity of the existence of publicly availability large BIM object databases, this MSc thesis will focus on the automatic generation of synthetic labelled point clouds of indoor environments from BIM models. These datasets will be further explored for classifying point clouds captured with Apple Smart Devices.

Required skills: Proficient in Python programming

Contacts

Keywords

BIM, Point clouds

Exploring the impact of tree original point cloud data leverage in urban daylight simulation 🔗

Description

Enhanced daylight exposure reduces reliance on artificial lighting, leading to energy savings and environmental benefits, while also contributing to healthier living and working conditions. In this way, urban daylight simulation is a key tool in fostering healthier, more sustainable urban communities. While there have been various studies on urban daylight analysis, the impact of urban trees is usually neglected, which can lead to considerable daylight overestimation. This is mainly caused by the lack of 3D tree data on neighbourhood or urban level. On the other hand, upon existence of such data, the complexity of 3D tree configuration leads to abstract/inaccurate 3D tree model reconstruction. 3D tree models, applied in simulations, are often reconstructed as polyhedral, which disregards the tree canopy porosity and leads to daylight underestimation. Furthermore, the seasonal alterations of trees can have significant impact on daylight simulation. In line with this, the employment of original point clouds in daylight simulations, rather than 3D reconstructed tree models, can provide a more accurate estimation on their impact on (urban) daylight simulations. In this regard, a potential thesis topic is the leverage of original point cloud data (instead of 3D tree model) of trees on neighbourhood/urban daylight simulation. The performance of tree point cloud data can be compared to reconstructed 3D tree models. Furthermore, different point cloud density can be considered in simulations, to represent the seasonal alterations of tree canopy densities and their impact on the daylight simulation.

Required Skill: Programming

Sources

Contacts

Keywords

Point clouds, Simulation

Extending AR point clouds to streets 🔗

Description

Context: Point clouds, as the third spatial data type in addition to rasters and vectors, are penetrating into nearly every spatial application. A fantastic part is to visualize them in reality, helping people comprehend the world more. Earlier work of our group demonstrates the possibility to show indoor point clouds in an AR environment, using only a cell phone. With this app, people can try lots of furniture in their rooms. We think the same idea could be applied to the outdoor environment as well, e.g., city planning, change detection, etc, to help professionals and laymen. However, the difference is that outdoor point clouds are usually much larger, which can cause critical issues when rendered on mobile devices. We have developed state-of-the-art continuous Level of Detail (cLoD) techniques, will they be helpful and applicable?

imgAR1.gif imgAR2.jpg

Source: images from Liyao Zhang (left) and the Internet (right)

Objective: The main research question is: What is an optimal cLoD for visualizing large outdoor point clouds on AR platforms? Sub-questions include: 1. What specific applications to consider? For example, adding virtual buildings or even a whole area; show changes occurred when scanning a certain area 2. Which parts need cLoD? Several bottlenecks exist when rendering large point clouds, e.g., data loading, GPU rendering, etc. 3. How to adapt a generic cLoD to an optimal solution, considering applications and bottlenecks? Our group has developed a generic cLoD framework. You can get first impression by using it. With your understanding of the problem, you will play around to improve the cLoD expression. By programming basic code and testing, such a project can surely be done.

Contacts

Keywords

Point clouds

Extracting Maritime Traffic Patterns for Energy Transition 🔗

Description

To create a sustainable world, we must minimize the human effect on producing Green House Gases (GHG). One of the main sources of GHG is transportation. Maritime transportation accounts for 90% of world trade, which presents a big opportunity for improvement. Energy Transition in the maritime sector can help achieve this improvement. However, this transition requires detailed and structured strategic and operational planning. Analyzing the present energy usage of vessels can help achieve this planning.

The energy usage of vessels depends on many parameters that come from different sources. Some parameters must be collected on board while others can be gathered from vessel's navigational information which can be reached from AIS data. AIS is a sequential point data broadcasted by vessels with varying frequency (2 seconds -- 10 minutes) depending on the manoeuvre characteristics of the vessels and message type. Some navigational features include Latitude, Longitude, Speed Over Ground (SOG), Course Over Ground (COG), MMSI (Maritime Mobile Service Identity). Besides, vessel information such as Size (Length Over All - LOA, Beam), Type and MMSI can also be gathered from AIS data. Analyzing AIS data for each ship's movement is computationally intensive and requires a proper spatial-temporal data organization for the purpose of the analysis.

maritime_energy1.png

maritime_energy2.png

Patterns along the routes extracted via geospatial analysis can be used for energy usage. To evaluate the energy demand of vessels, geospatial analysis focuses on the points where significant changes are observed in navigational characteristics. These points are called waypoints. There are three main techniques for waypoint analysis, each with its advantages and disadvantages:

In this project, you will research the possibilities for:

Machine learning algorithms may help in the extraction process and the outputs should be selected for hind-casting energy consumption of the vessels or energy demand along the routes. Throughout this study, the use of external databases for vessel engine type and size is highly recommended for increasing the accuracy of the energy demand prediction.

References

Contacts

Keywords

Database Management Systems

From low quality point clouds to high quality BIM 🔗

Description

The price reduction of LiDAR technology has allowed its integration into several devices not designed for surveys. This work will explore how the quality of data taken with low-cost devices (iPhone, HoloLens, and other HMLS devices) influences Scan to BIM. The robustness of the algorithms will be assessed considering variations in data quality related to accuracy, density and occlusions.

References:

Contacts

Keywords

BIM, Digital Twins, Point clouds

GeolinQ PC data management 🔗

Description

Contacts

Keywords

Point clouds

Geospatial - and Imagery Intelligence (GEOINT and IMINT), in Dutch 🔗

Description

military1.png military2.png

Big Data

Ethiek vs juridische grondslag: Binnen de militaire context een interessante casus. Ministerie van Defensie valt niet direct onder de nieuwe WIV (wet inlichtingen en veiligheid) zoals de AIVD, maar komt wel met regelmaat in aanraking met (persoons)vertrouwelijke gegevens. Als deze gegevens in grote getalen binnen gaan komen is het niet meer haalbaar om deze per individuele casus te beoordelen en hebben we dus richtlijnen nodig over wat wel en wat niet op te slaan. Dan begint niet alleen de vraag op te spelen wat we op mogen slaan, maar ook wat we op willen slaan en waar we de data voor mogen en/of willen gebruiken.

Tooling specifiek toegepast op bruikbaarheid voor onze analyses: We hebben uiteraard niet alle data die er binnen te halen valt nodig. Defensie analisten werken volgens een bepaalde methodiek, en programma’s/tools die ontworpen zijn om de voor hen bruikbare data zo efficiënt mogelijk te bevragen is nodig.

Algoritmes voor change detection bij onder andere grote hoeveelheden satellietbeelden.

Remote Sensing

Toepassingen voor remote sensing in de militaire context: Aan de ene kant het verfijnen en toepasbaar maken van de technieken om deze zo optimaal mogelijk in te kunnen zetten, aan de andere kant nadenken over de toepasbaarheid van deze technieken in uitzendgebied waar ICT ondersteuning regelmatig te wensen overlaat.

Databases

NoSQL database: Op dit moment werkt JISTARC op een relationele database, die met de komst van meer data tegen het maximale van zijn kunnen op gaat lopen. JISTARC wil overstappen naar een niet-relationele database, waarbij idealiter de vulling geautomatiseerd plaatsvindt. Bovendien moet er samen met de analisten worden nagedacht over de interface zodat deze alles bevat wat zij nodig hebben, maar niet te complex wordt.

Data virtualisatie/federalisatie: Hoewel JISTARC op dit moment niet altijd de systemen hebben om alles ook te testen, moet er nagedacht worden over de manieren van data virtualisatie/federalisatie en hypotheses worden opgesteld/onderzoek gedaan worden naar de beste aanpak/mogelijke implementatie.

Automatisering

Veel van de producten die JISTARC maakt hebben vaste formats. Per casus/operatie/gebied worden deze formats opnieuw gevuld. Het zou veel tijd schelen als grote gedeelten van dit proces geautomatiseerd zou verlopen. Denk hierbij aan het automatiseren van een ‘weather effect matrix’ (waarin wordt weergegeven hoe de weersomstandigheden invloed hebben op de verschillende aspecten van de militaire operatie) op basis van de meest actuele weersgegevens.

Contacts

Keywords

Military

Hierarchical topology structure and viewer in QGIS 🔗

Description

The project aims to investigate possible data structures for storing administrative subdivisions data (e.g. neighbourhood, municipality, province, country borders) consistently. One promising approach is using a 4D topological data structure stored in a DBMS. The subtopics of this project include how such data structure can be edited with QGIS, made consistent over updates over time (e.g. merge of municipalities from year to year) and visualized.

Contacts

Keywords

Database Management Systems

Historic development of the Dutch municipalities 🔗

Description

Municipalities form an important governmental structure in a country. Of course this is also the case for The Netherlands. Surprisingly enough there appears to be no authoritative administration for municipality-boundaries for this highly organized country. Several administrations do exist, but there are differences. There was aPhD study completed by 2007 on this theme ('Gemeentegrenzen in Nederland: Een juridisch, technisch en kadastraal onderzoek'): http://repository.tudelft.nl/islandora/object/uuid:af284777-3f6e-4fb5-8b85-f4a6f4c2c758?collection=research (pdf download on the bottom, Dutch only). To complicate matters, there is a yearly pace of changes in municipalities, usually merges. Over the years some 1700 municipalities have existed, of which currently (2017) only 388 remain: https://www.cbs.nl/nl-nl/onze-diensten/methoden/classificaties/overig/gemeentelijke-indelingen/indeling%20per%20jaar/gemeentelijke-indeling-op-1-januari-2017

mun1.gif mun2.png mun3.png

Figure. left: number of municipalities, middle: map 1840, right: map 2009 Images taken from http://members.ziggo.nl/ekamper/gemeenten/gemgrens.html

The various administrations are, as usual nowadays, scattered over the Web (and probably other places) and sometimes fragmented / with limited access. It would be of public interest to build a comprehensive dataset containing the various municipality data, including geometry and offer both an API and a visualization. The upcoming DSO ("Digitaal Stelsel Omgevingswet", https://www.omgevingswetportaal.nl/ ) could be one of the main 'users'. From a data-point of view there are two interesting dimensions in the municipality data: the time dimension (valid time only) and the "reality" or the administration. A private product-in-development,EntityGrid , claims to be able to handle these dimensions in data for registration and easy access by one or more APIs.EntityGrid is built in Java and uses an In-Memory Data Grid (IMDG) for persistence. This subject could handle the following aspects: • make an inventory of data sources about municipalities and their geometries over time • design a spatio-temporal-thematic information model for this registry • gather the historic data and build a temporal and multi-reality dataset in [ EntityGrid • develop approach for maintaining this registry • create APIs to access the dataset using both the time dimension and the reality-dimension possible forms: o Java API o REST API o Linked Data API • create an online visualization using both the time dimension and the reality-dimension

Some links:

More links and raw data available.

EntityGrid is a product developed by Rob van Dort. He has developed an initial text-based demo ( http://145.131.1.55/munidemo/2202c6e ), he will assist during the realization of this case.

Contacts

Keywords

Database Management Systems

Identifying blockchain implementation for 3D Land Administration in line with ISO 19152 LADM Edition II 🔗

Description

This thesis will explore the potential of integrating blockchain technology into land administration systems, with a specific focus on aligning this integration with ISO 19152 Land Administration Domain Model (LADM) and specifically ISO19152-6: Implementation. The research aims to develop a blockchain-based framework that enhances the security, transparency, and efficiency of land administration processes. The potential benefits and challenges of using blockchain technology in land administration will be addressed, focusing on the technical challenges, following the investigation of front-runners in this domain.

Contacts

Keywords

Blockchain, Database Management Systems, Land Administration

Identifying land-related indicators for ISO 19152 LADM Edition II 🔗

Description

The 2030 Agenda for Sustainable Development is a global plan of action for people, planet, prosperity, peace, and partnership. UN member States endorsed the 2030 Agenda and committed to implement the Sustainable Development Goals (SDGs), a set of 17 Global Goals in a 15-year period. Among these goals, some are intricately tied to the realm of land, emphasising the critical importance of effective land management and equitable land distribution for sustainable development. Moreover, the Global Land Indicator Initiative (GLII), seeks to derive a list of globally comparable harmonized land indicators, using existing monitoring mechanisms, and data collection methods as a foundation.

Land administration relates people to land and informs on the 'how', the 'what', the 'who', the 'when' and the 'where' of land tenure, land use, land value, and land development. Land Administration Systems (LAS) are the basis for recording the complex range of rights, restrictions and responsibilities related to people, policies and places. LAS based on ISO 19152 LADM, are key drivers towards a sustainable economy and society and the Edition II of the standard, which is currently under development, will be key component alongside agreed global concepts and evidence-based approaches. Therefore, there is a need for a foundation of a Land Administration Performance Index -- which is a possible link to existing global frameworks or initiatives.

Objective: This MSc thesis will investigate the identification of the ISO 19152 LADM - related SDGs, specifically SDG 1, 2, 11, 14 and 15, provide a formal way to express them within LADM and explore visualisation options (alternatives to visualise them could be: story maps, National Tenure Atlas, etc.). How argets and indicators can be used to assess the added value of adopting 3D Land Administration Systems based on LADM or provide insights if a country does not adopt 3D LAS, will be explored.>

3D_Cad2.jpg

Related literature

Contacts

Keywords

Database Management Systems, Land Administration, Sustainable Development Goals

Improved cadastral map updates 🔗

Description

The Dutch cadastre is investigating the possibility to improve the cadastral map in two ways: adding quality description to the map elements and improving the geometric quality itself by recalculating the map with data extracted from field documents. A prototype for a semi-automatic reading of the old documents in being developed, connecting the survey data from different source documents and integrating them together in order to build up a very accurate geometry. From this geometry the new map has to be extracted.

The described production software is currently being developed. One of the main goals at this stage is to create a very rich data model which makes it possible to produce solutions with the highest degree of automation. This data model is aimed to be on the cutting edge of GIS (the current cadastral map) and geodesy (measurements & points). Possibly subjects of interest will be:

Special attention must be paid to the survey and representation package of LADM and the adjustment policy that can be used to recalculate the map, especially in light of the refined survey model that is being developed within the scope of the LADM revision.

Furthermore, an innovative and unconventional approach for calculations could be explored for computation and enhancement of the accuracy, such as Genetic Algorithms. This biological optimization technique has proven to be very effective in solving problems with multiple parameters and constraints, as in the case of cadastral maps and various survey data.

image001.jpg

Contacts

Keywords

Land Administration

Improving the content of vario-scale maps 🔗

Description

The main goal of this research is to improve the automated generalization process for vario-scale maps at non-fixed levels of detail (stored with the tGAP data structure).

gen_operations.png

There are lots of possibilities for research and improvement:

A possible starting point for this research is the following paper, that describes some initial ideas for orchestrating the generalization process, but has not yet been implemented so far.

Contacts

Keywords

Vario-scale

Indoor/Outdoor seamless modelling intended for navigation 🔗

Description

IndoorGML is an Open Geospatial Consortium (OGC) standard that focuses on the representation and exchange of data representing the 2D and 3D indoor environment. It provides a geometric, topological, and semantic description of the indoor which facilitates specifically applications like indoor navigation or facility management. While other standards dealing with indoor (e.g., IFC,CityGML ) represent the structural elements of buildings, they lack the representation of spaces and their topology, and consequently, geospatial applications such as navigation is not immediate. This MSc thesis will focus on the development of a method for the integration ofIndoorGML with outdoor networks in order to enable indoor/outdoor seamless modelling,

Required skills: Proficient in Python programming

References: https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLIII-B4-2022/295/2022/

Contacts

Integrating point cloud data sets at multiple epochs: A higher point density and filling occlusions 🔗

Description

The goal of this project is to investigate reliable and verifiable techniques for integrating point cloud data from multiple acquisition moments (epochs) to enhance point density and address gaps in the data caused by the acquisition process (occlusion). This challenge is similar to change detection [2], but also involves identifying obscured areas to create a more complete point cloud representation. One potential solution is to analyze the trajectory of the lidar rays from different epochs using tools likeOctoMap [1]. For this, you need to know the sensor position, next to the points that are measured [3]. A case study using AHN1, 2, 3, 4 and 5 or data collected with our own handheld lidar scanner (Geoslam Zeb Horizon) should be included.

church_tower_occluded.png

References

  1. https://link.springer.com/article/10.1007/s10514-012-9321-0
  2. https://doi.org/10.1016/j.isprsjprs.2023.01.010
  3. https://eartharxiv.org/repository/view/119/

Contacts

Keywords

Point clouds

Integration of point clouds in CityGML standard for obtaining a continuous Level of Detail 🔗

Description

Context: One of the main characteristics of the CityGML standard is the consideration of different Levels of Detail (LoD0 to LoD4 ) [1]. The LoDs allow discrete visualizations of buildings and urban infrastructures. The original five LoDs have been extended with four sub-levels [2]. However, LoDs of CityGML are still far from being considered continuous LoD (cLoD), as in point clouds or maps, where recent works have shown the benefits of cLoD in improving user interactions [3].

CityGMLcLOD.png

Objective: The aim of this work is to study the integration of point cloud geometry in the CityGML standard to obtain a cLoD, similar to the existing in point clouds and reaching at discrete levels the traditional content.

References:

Contacts

Keywords

Point clouds

Intelligent routes with social distancing 🔗

Description

Context: In https://www.metronieuws.nl/in-het-nieuws/2020/04/dringende-oproep-vier-koningsdag-thuis it was expressed that: "Hubert Bruls, wil kijken of er slimme looproutes in stadscentra kunnen komen, vooral op smalle plekken waar mensen moeilijk voldoende afstand kunnen houden." or "Hubert Bruls, wants to see if smart walking routes can be created in city centers, especially in narrow places where people find it difficult to keep a sufficient distance." Hubert Bruls, is the "voorzitter van het Veiligheidsberaad van burgemeesters" ("Chairman of the Mayors Security Council") and this statement was made in context of the corona virus..

anp_utrecht2019.jpg

(photo by ANP, Utrecht 2019)

Objective: In this research several aspects have to be covered: creating data model for storage of this information (including the navigation graph and the geometry of the street), collecting data to populate the model (at least sample data for research), computing the intelligent routes with social distancing, and (optionally) a nice user interface on a mobile phone. Aspects to be covered in path planning, by assigning to navigation graph: 1. geometry only (width of the street) 2. statistics on crowded locations based on historic data 3. real-time data on crowded locations 4. in case of 3, also update route during walking (if crowded situation changes), i.e. dynamic routing

It is a bit related to the earlier routing research projects at TU Delft, such as fire fighter routing, avoiding dangerous areas, in form of moving plumes or forest forest, that evolve and may block roads, and pedestrian navigation; see:

Contacts

Keywords

Database Management Systems

Interpreting spoken text to a map (at CGI) 🔗

Description

Spoken text can be digitized using Large Language Models (LLM's), like LLAMA. With some additions CGI is capable of finding absolute location pointers in those digital texts. So we can find locations in spoken text. The next step is to interpret these pointers into a map. For that we will also need to find the relative pointers (like "to the right of", "in front of", "opposite of") and directional pointers (like "from A to B"). And all these pointers (absolute and relative) need to be fused into a geographical description (i.e. a map). Some iteration might also be needed here.

The benefits for society are that with a method like this we can interpret messages from ordinary people who communicate with first responders in a geographical way, which makes it much easier for the people in the call center to plan and communicate this plan to their operatives.

What we (CGI Nederland, location Rotterdam) are looking for is a student, that wants to build a first proof of concept of this whole method (extending a first Proof of Concept by CGI). Dealing with the relative and directional pointers is the most difficult part and will need some significant coding to be done. Python can be used and in this first PoC we will stick to 2D. But also a further use of the Speech to Text part for finding a longer list of absolute and relative pointers is needed, so LLM's will also be used (and potentially trained).

Interpreting spoken text to a map

Contacts

Keywords

Artificial Intelligence, Georeferencing

Investigating the reliability of indoor objects to act as landmarks 🔗

Description

Landmarks play a critical role in user guidance. Landmarks are used in several cases: 1) to understand current location ( i.e. where I am); 2) to help user distinguish directions ( i.e. where I will head); 3) to confirm user on the right track. At present, the research of navigation landmark for pedestrian focuses on outdoor environments. A landmark could be a high-rise building, a crossing, a signage or any other salient objects to pedestrians. Indoor landmarks are studied in robot navigation. However, there are seldom studies on indoor landmarks for pedestrians.

Indoor landmarks can be a 'coffee machine', a 'picture on the wall', 'plant', etc . If pedestrians are given indoor landmarks, they can easily find their way especially in large halls (convention centres, museums, airports, etc .). Yet indoor landmarks may be distinct to different groups of people (classified by gender, age, height, ability, etc. ).

The goal of this research is investigating and classifying different indoor objects which can be used as landmarks for different types of pedestrians. Some of the research questions will be: "how many user groups can be identified?", "how the landmarks can be organised in 3D indoor navigation models?", "what is the reliability of these landmarks?", "is it possible to obtain these landmarks automatically?".

Contacts

Labeling vario-scale maps 🔗

Description

Icons and text are an important part of every map. Symbols and labels have not been implemented in the automated generalization process for vario-scale maps. Placing labels is a classical cartography problem: Each object that has to be labeled allows a number of positions where the corresponding label can be placed. However, each of these label candidates could intersect with label candidates of other objects. Optimization approaches exist that allow to find a solution for just one predefined selected scale, but how can we find the right solution for vario-scale maps? The main goal of this project is answer this and other questing and investigate options how we can produce, store and visualize good labels and icons in vario-scale maps. A student should explore current solutions for label placement in dynamic maps environment such as [4] and Mapbox [5], where the navigation is the main purpose and for symbols and labels is crucial to avoid behaviour that is distracting or jarring such as labels popping or moving about in unexpected ways [2]. Possible research directions: Will the labels be placed on the static map retrieved from the vario-scale structure? For example, the map is created one slice of smooth tGAP structure and labels are added additionally on the fly similar to [3]? Or will the labels be encoded in the structure? E.g. every object in the structure should have the label placed in its area. The size of a label could be defined by the boundaries of the area objects. This idea originates from [1] and it can result in a situation depicted in the figure below.

labelling_varioscale.png

Using and testing with real GIS data is an essential part of the thesis. A result should preferably also be integrated into the builder for constructing a vario-scale data structure.

Contacts

Keywords

Vario-scale

Making a vario-scale 4D point cloud 🔗

Description

Current point cloud visualizations often employ octree structures for managing Levels of Detail. This leads to visible artifacts, while interacting with the point cloud. In this thesis you will investigate to what extent it is possible to add continuous level of detail information as a 4th dimension to the point cloud (i.e. 1 added attribute).

discrete_lod.png

This would mean creating a point cloud with 4 dimensions, and then using this 4th dimension, while interacting with the point cloud in an interactive way to only display relevant points for the subset of space displayed to the user. A smooth point cloud viewer that runs on the web would be one of the outcomes of this research. Hence, the topic would require a thorough background in, or a strong desire to learn, javascript &WebGL (and related technologies).

Contacts

Keywords

Point clouds, Vario-scale

Matching the Dutch Cadastral Information Model with the ISO Land Administration Domain Model (at Kadaster) 🔗

Description

The ISO standard for the Land Administration Domain Model has been accepted in 2012, and by now many countries have published their LADM country profile, and several countries also used LADM as basis for the renewal of their information system [1]. The use of LADM is encouraged, by UN-GGIM, UN-Habitat, and the World bank. Currently, the LADM revision process is on-going, based on new requirements from ISO member states. This is resulting in a multi-part Edition II of LADM, covering: land tenure, marine georegulation, land value, and spatial plan information [2]. In parallel the Dutch Cadastral Information Model (IMKAD) has been evolving independently the past decades to serve the needs of the Dutch society [3]. There have been some applications of LADM in Netherlands; e.g. for supporting Cadastral Map Renovation [4], for a Linked-data standards-based portal for integrated Land Administration information [5], and when making a valuation Information Country profile [6]. Note: the information models of all ISO TC211 standards, including LADM (ISO 19152 series), are available from the GitHub page of the Harmonized Model Maintenance Group [7].

Figure 1 - Simplified representation of the LADM Model
Figure 1 - Simplified representation of the LADM Model

Figure 2 - Simplified representation of the IMKAD Model
Figure 2 - Simplified representation of the IMKAD Model

It is now a good moment to reflect and compare the two information models. Is the scope/intention of both information models comparable, with LADM Edition II this was extended (marine, valuation and spatial plan information)?, How are the persons/parties, the Rights, Restrictions, and Responsibilities (RRRs) for the various types of spatial units/parcels (land, apartment rights, legals spaces utilities, legal and spatial sources, etc. represented in IMKAD and LADM?, Would the basic administrative unit (property unit), also be valuable for The Netherlands?, What about 2D/3D, geometry/topology, (bi-)temporal aspects (real world vs. database/registration time), versioning (at object or attribute level), etc.? What about group rights, informal rights, community rights, conflicting/overlapping rights (e.g. based on historic proof but in contrast with today's use/registration)? What about the content of the various code list, i.e. the actual code list language?, And last but not least, as the languages are different, how should this be handled?

Research topics

Next to the questions mentioned in the description above, below an initial list of research topics/ question/ aspects to investigate:

References

Contacts

Keywords

Data transformation, Land Administration, Model mapping

Mobility of People 🔗

Description

There are loads of datasets with tracks of people collected by GNSS collected by different groups, for example pedestrians in city centres or even more interesting all mobility of people in one week. This research intends to investigate rules for mobility per neighbourhood, i.e. defining a relation between spatial configuration of programme and network, position in the city, position in the region and the actual movement pattern of people (mode, destination). The research should also investigate which other spatial data sets will be needed and connect those to the available mobility data.

Contacts

Nationwide geo-referenced repository of BIM/IFC 🔗

Description

Designing and testing an architecture for the exchange of an nationwide geo-referenced repository of of BIM/IFC models, assuming storage in a spatial Database Management Systems and dissemination via web-services (both download and viewing services).

Contacts

Keywords

BIM

Netatmo Sensor Integration in IFC model of TU Delft Architecture and the Built Environment building 🔗

Description

Monitoring building performance on different aspects, such as temperature, noise and CO2 level is essential for gaining insight on the current situation aiming at improving living conditions on individual building level as well as mitigating environmental problems on larger scales. Through the advancements in Artificial Intelligence, data-driven approaches for building performance monitoring and predictions are being increasingly applied. Various sensors are available and can be integrated in buildings for collecting data on temperature, noise, CO2, humidity, etc. Efficient incorporation of such real-time and historical sensor data into Building Information Models (BIM) supports their usability in various analyses, including the connection to building physical/geometrical properties, for instance in building energy performance analysis. On the other hand, optimal placement of sensors in a building is crucial for exploring the impact of building physical/geometrical properties on its performance (e.g. temperature). This research topic consists of two parts. The first part of the research comprises a research on the optimal placement of the available 7 Netatmo indoor sensors in Architecture and the Built Environment building rooms so that it provides the scope for further research on building performance (e.g. on temperature). The second part of the research focuses on the integration of these (real-time) sensor data (requested through Netatmo API) into a BIM (IFC) model of Architecture and the Built Environment building. Furthermore, the BIM-integrated Netatmo sensor data will be incorporated into a 3D web platform, such as Cesium, wherein real-time and historical data of each room can be visualized. Particular attention should be paid to how the data needs to be stored and which real-time web protocols are available and can be employed for this purpose, like MQTT and websocket.

Netatmo.PNG

Contacts

Keywords

Geoweb

OGC API - Processes for the dissemination of simulation results 🔗

Description

Environmental modelling often requires multiple simulation models and (heterogeneous) data from different sources. Coupling these simulation models and data sources, originating from different tools (and data providers) which follow different protocols can be challenging. A service-oriented approach which provides a standard interface for the interoperability between the different simulation models and their corresponding data sources support a more seamless integration of different simulation models and their incorporation into different systems. OGC API- Processes provides a standard interface for making computational geospatial processing services accessible, via web services. OGC API- Processes leverages the concepts from OGC Web Processing Service (WPS) 2.0 Interface Standard, but is a more modern way of interacting with (computational) resources over web, communicating over a RESTful protocol usingJavaScript Object Notation (JSON) encodings. This research topic can focus on an efficient framework design and implementation of disseminating urban simulation models (e.g. noise, wind) through OGC API -- Processes standard over the web. These simulation models can be visualized in a web client, such as Cesium, containing 3D city models. These 3D city objects will also be applied as input of the simulation models conform OGC API -- Processes standard. Furthermore, the potency of interoperability between the simulation outcome of the involving simulation models can be explored.

Processes.PNG

Contacts

Keywords

Geoweb

Optimization for energy demand-production-storage balance in Local Energy Systems through Deep Reinforcement Learning 🔗

Description

Local Energy Systems (LES) play increasingly significant role in energy transition, focusing on generating, storing, and distributing energy within a localized area. These systems enhance energy efficiency, reliability, and sustainability by integrating local renewable energy sources like solar and wind, thereby reducing dependence on centralized power grids. To realize the minimum dependence on the centralized grid, local energy systems should facilitate the balance between energy demand and energy production. This can be realized through an optimized time-dependent energy distribution within the local energy system. For instance, buildings with different functionalities (e.g. residential and offices), having different demand patterns (residential buildings have mainly higher demands in the evening, while offices during the day) can share energy at different hours of a day. Likewise, the energy production of renewable sources (e.g. solar) is time-dependent. Therefore, a time-dependent optimized solution for energy distribution between the different entities of the local system, namely sinks (building energy demand), sources (renewable energy generation) and storage (batteries and electrical vehicles) increases the efficiency of the Local Energy System and minimizes the dependence on the main grid. A potential thesis topic is development and implementation of Deep Reinforcement Learning approach upon an existing Geomatics MSc thesis research. While this existing research has mainly focused on the leverage of Deep Reinforcement Learning approach for a time-dependent solution on a neighbourhood level, the new research can further expand the Deep Reinforcement Learning, through the incorporation of a graph-based approach, for the development of a time-dependent optimized solution for energy distribution between the individual entities of the Local Energy System.

Required Skill: Programming

Local Energy Systems (LES)

Contacts

Keywords

Artificial Intelligence, Digital Twins

Optimizing Solar Panel Installations in Urban Districts: A Data-Driven and Digital Twin Integrated Approach 🔗

Description

Background and Aim: In the face of the global shift towards decentralization, decarbonization, and digitalization in energy systems, the importance of solar energy as a key renewable energy source (RES) has become increasingly prominent. This research aims to forge a path for advanced decision-making in urban solar installations by creating realistic planning scenarios that consider sustainability goals and renewable energy uptake. By leveraging data layers, spatial analysis, and optimization algorithms, this project will not only explore the potential for solar energy generation in urban districts but also optimize the siting of installations based on various parameters and energy demands. Additionally, the integration with a web-based digital twin model will enhance visualization and the practical application of findings in real-world settings.

Research Questions:

  1. What methodologies can be employed to analyze and enhance the deployment strategy of solar panels in urban districts and buildings using current and historical data?

  2. In what ways can optimization algorithms be applied to determine optimal solar panel placements within urban landscapes?

Research Objectives: This research seeks to develop a robust framework for the strategic planning and optimization of solar panel installations in urban areas, which includes:

  1. Mapping out current and historical data to identify trends and gaps in solar panel installations.

  2. Creating a multi-layered data integration strategy that incorporates energy demand, environmental impacts, urban layout, and infrastructural constraints.

  3. Applying advanced optimization algorithms to propose efficient and feasible solar installation scenarios.

  4. Integrating research findings with a digital twin model to provide a dynamic, interactive visualization tool that aids in decision-making and scenario testing.

Method:

The methodology encompasses a mix of theoretical research, practical data analysis, and technological application, as follows:

PS:

Significant progress has already been made in the development of the necessary data and the digital twin model. Assistance with integrating your model into the existing digital twin framework will be provided by Amin. You can view a demo of the digital twin model here: https://dataless.beta.geodan.nl/.

References: [1-4]

  1. An, Y., et al., Solar energy potential using GIS-based urban residential environmental data: A case study of Shenzhen, China. Sustainable Cities and Society, 2023. 93: p. 104547.
  2. Gassar, A.A.A. and S.H. Cha, Review of geographic information systems-based rooftop solar photovoltaic potential estimation approaches at urban scales. Applied Energy, 2021. 291: p. 116817.
  3. Caprari, G., et al., Digital Twin for Urban Planning in the Green Deal Era: A State of the Art and Future Perspectives. Sustainability, 2022. 14(10): p. 6263.
  4. Narjabadifam, N., et al., Framework for Mapping and Optimizing the Solar Rooftop Potential of Buildings in Urban Systems. Energies, 2022. 15(5): p. 1738.

Contacts

Keywords

Digital Twins, Energy, Optimization

Patterns of room occupancy in houses and their relation with energy use 🔗

Description

Occupancy behavior and patterns of room usage are believed to influence the energy consumption in homes. Unfortunately, little is known about these behavioral patterns. Our group started a few years ago to build up a database of houses in which occupancy, comfort, temperatures, indoor air quality, gas and electricity use are measured for long periods with a frequency of 10 minutes. At the moment our database contains around 100 houses. This master thesis is about finding patterns of room occupancy at house level and studying possible relationships with energy use. Finding ways for easy-to-read graphical representation will be needed as well. The master thesis is part of the OPSCHALER project, in which a PhD student is also working (www.opschaler.nl). The thesis should answer the following questions:

a) Which combination of datasets (movement sensors, CO2 data) makes it possible to determine the presence in rooms?

b) Which rooms are used all over the day, in which sequences and is there a specific pattern of use in each household?

c) How can this patterns be graphically represented?

d) Is there similarity of patterns throughout the households?

e) Are there correlations between room occupancy and electricity or gas use patterns?

Contacts

Keywords

Digital Twins, Energy

Point clouds in the cloud: MS Azure, Amazon, etc 🔗

Description

Contacts

Keywords

Point clouds

Positioning and Location Awareness, Point Cloud Cartography or Point Cloud Analysis 🔗

Description

Any topic related to:

  1. GEO1003 Positioning and Location Awareness
  2. Point Cloud Cartography
  3. Point Cloud Analysis

If any of these topics do have your interest, contact Edward Verbree, discuss and agree with him on a topic.

Contacts

Keywords

Gaussian Splatting, Point clouds, Positioning, Visualisation

Possibilities of using Solid data pods in a geo environment 🔗

Description

In this research proposal we wish to research the possibilities of a Solid data pod in the context of geospatial data. Solid is a web decentralization project led by Sir Tim Berners-Lee. The project aims to radically change the way Web applications work today, resulting in true data ownership as well as improved privacy by developing a platform for linked-data applications that are completely decentralized and fully under users' control rather than controlled by other entities. Solid is a specification that lets people store their data securely in decentralized data stores called Pods. Pods are like secure personal web servers for data. When data is stored in someone's Pod, they control which people and applications can access it. All data in a Solid Pod is stored and accessed using standard, open, and interoperable data formats and protocols. Solid uses a common, shared way of describing things and their relationships to one another that different applications can understand. This gives Solid the unique ability to allow different applications to work with the same data.

Possible research questions

In this research the following steps can be taken:

  1. Literature research of Solid and applicable geo-standards.
  2. Description of one or more use-cases for spatial Solid.
  3. Implementation of a use-case in a working environment.

For the implementation of the use case we can make use of the Flanders Solid platform .

Suggested use cases

Sharing location history

Mobile phones can record the location of the phone. This location could be shared in many different ways: live location sharing is possible, but also recording a location or trajectory for sharing at a later date. Also, there are possibilities to enhance the location with specific attributes. or example, an organization that wishes to make a map for inline-skaters. For inline-skating you need a very smooth road. A skater may install an application that not only records the GPS track but also enriches the track with extra properties of the track entered by the skater, such as the road condition and surface material.

Research Questions:

Sharing species distribution

The zandmotor ( https://dezandmotor.nl/ ) is a project where a whole new piece of land is created on the Dutch coast. Ecologists are interested in the development of plant life on this new piece of land. As the land is publicly accessible people may contribute to the research by submitting observations via apps likeObsIdentify ( https://observation.org/apps/obsidentify/ ). Is there also a possibility to use a data hub for this kind of data sharing

Research Questions:

Contacts

Keywords

Geoweb

Predictive fetching of geo-web resources 🔗

Description

While browsing interactive maps, data needs to be shipped from server to client for rendering. There exist many standards making this possible, e.g. WMTS, I3S, etc. Most communication schemes for data retrieval are used reactively, meaning that first the user has to carry out a user action like zooming or panning, before data is retrieved. In this project you will study the possibilities for pre-dictive fetching of resources. You should build a model to predict user actions and pro-actively retrieve resources before they are actually needed. The model should be implemented and tested, e.g. by including it into the vario-scale web viewer: https://varioscale.bk.tudelft.nl/gpudemo/2019/12/drenthe/

Relevant literature:

Contacts

Keywords

Geoweb

Rights in multi-functional buildings 🔗

Description

Including future ownership and use rights during the design of a (multi-functional) building complex, by enriching the BIM/IFC spaces with legal boundaries, identifiers of legal spaces, and proper georeferencing (design information model, and assess with a number of buildings)

Contacts

Keywords

Land Administration

Scan2BIM 🔗

Description

Scan2BIM , using laser-scanning, optical imagery enrich the resulting indoor/outdoor pointclouds not only geometrically (georeferencing, continuous levels of detail/importance), but also enrich them semantically by using classifications according to BIM/IFC (i.e. the IFC files use pointclouds only to represent the geometric primitives).

Contacts

Keywords

BIM, Point clouds

Seabed fingerprinting 🔗

Description

Seabed fingerprinting, how do you store characteristics of the seabed in such a way that you can determine where you are based on this previously stored dataset (fingerprints) using inertial navigation and compass function, even without GNSS.

Contacts

Seamless Oblique Image Mosaics for Aerial Visualization (at Geodelta) 🔗

Description

This research project (graduation internship at Geodelta) focuses on creating seamless mosaics from oblique aerial photographs. The project aims to develop techniques to stitch these oblique photos together, allowing for a continuous and immersive visual experience when navigating through the images.

The objectives and challenges of this project include:

  1. Seamless Image Stitching: Developing methods to stitch overlapping oblique photos along the flight path, ensuring a smooth transition between consecutive images with minimal visual artifacts, and able to handle different amount of overlap.
  2. Enhanced Aerial Visualization: Create a viewing system that allows users to fly through a series of stitched oblique images without leaving the image view, providing a continuous visual experience, while maintaining the accuracy of the stitched images in relation to the actual 3D environment.

By developing techniques for seamless oblique image mosaics, this project will enhance the capability of visualizing aerial imagery, benefiting applications such as urban planning and environmental monitoring.

Seamless Oblique Image Mosaics

Geodelta is looking for a motivated student who is willing to make a deep-dive in photogrammetry and software engineering. While we are happy to provide a lot of support, the following competences will give you a head-start in your internship:

At Geodelta, we know that measuring is more than just precision. It is the reliability that counts. Are the measurements correct and have the correct procedures been applied? We are engineering consultants focussed on delivering the best technical advice and software in the fields of geodesy, photogrammetry and laser scanning. We make sure that our clients can trust their measurements. We do this by developing user-friendly software and by providing trusted advice. We are independent, innovative, pragmatic and above all we have a passion for geodesy. We are based in the century-old former faculty building for Geodesy. This is just around the corner of the building of the faculty of Architecture. If you are interested in this topic, or if you have an amazing idea yourself, let us know! We will arrange meetings with interested students discuss the project(s) and find the best fit.

Contacts

Keywords

Imagery

Sensor web / Internet of Things 🔗

Description

Contacts

Keywords

Geoweb

Showing Point Clouds in VR 🔗

Description

point_clouds_in_VR.png

A critical issue for massive point clouds rendering in VR lies in the limited memory capacity of the headset. Large scenes built by points are not available unless effective generalization techniques are applied. Besides, extracting points in the view from the database would be expensive if the scene keeps refreshing all the time. This could then influence the smoothness of VR rendering. To address these problems, you will study and develop specific Level of Detail (LoD ) and caching strategies of point clouds that facilitate scene construction in VR. In fact, we have already developed an efficient data structure for point clouds management and querying. You could then integrate novel ideas into it to realize smooth and efficient point clouds rendering. Programming skills are essential for this project.

Contacts

Keywords

Point clouds

Smart Point Clouds of heritage objects 🔗

Description

Creating and dissemination Smart Point Clouds of heritage objects, design the workflow, develop the processing, storage (DBMS), dissemination (Webservices) and visualiaztion (WebGL /WebVR), and test the proposed solution with 2 or 3 case studies.

Contacts

Keywords

Point clouds

Smart voxels 🔗

Description

Smart voxels, how can you make them represent interior spaces as best as possible, by not seeing the voxels as a representation of a point cloud, but as a representation of the physical boundaries and openings.

Contacts

Solar envelope zoning in dense urban environment 🔗

Description

Background and aim: The 3D Solar Envelope can be perceived as a tool for urban planning in dense built environment ensuring that adjacent neighbours have a specified minimum direct solar access time per day throughout the year. In this scene, the Solar Envelope Zoning is an approach to solar access protection, originally conceived as a legal framework for city development in the same sense as traditional zoning plans. Given the advances at the 3D city models, parametric CAD environments, digital twins and digital urban planning it is worth exploring how solar envelope can be efficiently defined for dense urban environments in order to used in various phases of the Spatial Development Lifecycle, facilitating its wider use as a prescriptive zoning tool.

Research questions: How can 3D solar envelope be modelled to provide zoning restrictions in the context of the Spatial Development Lifecycle (SDL)?

solar.png

Research objective: This research aims to investigate how the zoning restrictions from solar envelopes can be modelled in a structured way so that they can be used across various phases of the SDL. To reach this aim, the following steps need to be undertaken: 1) Identify the design methods and digital tools to define solar envelops in 3D. 2) Investigate how the zoning restrictions of those solar envelopes can be modelled according to ISO19152:2012 LADM.

Method: Literature research; Investigate potential application areas; develop and validate creative design solutions; elaboration of a case study.

References:

Contacts

Keywords

Land Administration

Spatial constraint modelling (3D geo-OCL, IFC/CityGML /IndoorGML ) 🔗

Description

Contacts

Spatial Resolution Enhancement of Thermal Satellite Imagery through Deep Neural Networks 🔗

Description

Thermal satellite imagery, records the emitted radiation in the Thermal Infrared region of EM spectrum, through which the temperature of objects on the Earth's surface can be measured. Thermal satellite imagery can be used for a wide range of applications, including climate monitoring, environmental studies and urban planning (e.g. in Urban Heat Islands and urban energy efficiency analysis). However, the native spatial resolution of these images is often limited by the sensor's capabilities (i.e. aperture size due to TIR long wavelength nature), leading to coarse and less detailed representations, which limits their application on more detailed scales. Therefore, enhancing spatial resolution of thermal imagery can improve their usability in various studies. Recent developments in Deep Neural Networks (DNN) for imagery spatial resolution enhancement, for instance, super-resolution generative adversarial network (SRGAN), demonstrated their potency. However, the focus lies mainly on optical imagery. In line with this, a potential thesis topic is the research and implementation of a DNN approach (e.g. an SRGAN approach) for the spatial resolution enhancement of thermal satellite imagery. Required Skill: Programming

Spatial resolution enhancement method for Landsat imagery using a Generative Adversarial Network

Source: Pham, V. D., & Bui, Q. T. (2021). Spatial resolution enhancement method for Landsat imagery using a Generative Adversarial Network. Remote Sensing Letters, 12(7), 654-665.

Contacts

Keywords

Artificial Intelligence, Imagery

Storing AIS track logs in a DBMS 🔗

Description

All larger ships (with length greater than 20 meters) are obliged to carry an AIS transponder that transmits their position to nearby ships. There are many applications for which this data may be used. A Database Management System (DBMS) will be needed to serve querying and archiving needs. Because of the amount of data (especially because historic data also needs to be stored), and the real-time nature of the dataset a sound design of a structure inside a DBMS is needed.

ais.png

In this project you will research the possibilities for building such a system and study one or more of below problems:

For this assignment a rather large archive of AIS messages is available (300GB). You likely will program in Python (or another language of your choice).

Contacts

Keywords

Database Management Systems

Straight skeletons for vario-scale maps 🔗

Description

It is known that Straight Skeletons can be used for map generalization [1]. Also it can be used to generate offset curves (buffering) and therefore it seems suitable for erosion/dilation of geometry (first buffering certain distance inwards then same distance outwards hence loosing unimportant parts of shapes). However, the straigt skeleton offers slightly different properties compared to normal buffering operations as implemented in mainstream GISes.

blow-shrink.png

In this project a student should therefore investigate what we can do with the Straight Skeleton as supporting data structure for vario-scale map generalization: how does the operation fit with generating smooth transitions for vario-scale map generalization, for which feature class types is it suitable and how does it compare with other alternative implementations that try to achieve similar goals?

Possible research directions: Can we generate smooth transitions for smooth zooming? Erosion / dilation (what is good order: first in, then out, vice versa, cf. [4]) and how to make these smooth? For which feature class types can this operation be applied, e.g. houses / built-up areas / archipelago / ...?

Investigations with some larger samples of real world geo-data (e.g. BAG panden) is a mandatory part of the project. To get started a student could use either Grassfire [0], which implements Straight Skeleton construction similar to [2], CGAL [3], campskeleton [6] or Surfer2 [7]. For comparison with other approaches one could consider the GEOS buffer operation (as exposed in e.g.PostGIS /Shapely) or using Clipper [5]. A result should preferably also be integrated into the builder for constructing a vario-scale data structure.

Contacts

Keywords

Vario-scale

Surface reconstruction from ALS point clouds by deriving isocontours 🔗

Description

For many applications (noise, wind and flood simulation models) a water tight surface representation is expected as input. In this project you will research a possible way to derive such a surface.

Starting point is deriving isocontours from an airborneLiDAR scanned point cloud (i.e. AHN3) by using a Delaunay triangulation and the chi-shape algorithm [1] or an alpha shape algortihm. This leads to a stack of polygons (possibly with holes), the iso-contours.

isocontours1.png isocontours2.png

The first objective in the research is to find a way to get from the iso-contours to a surface representation. A starting point could be [2], who employ the straight skeleton to obtain an interpolation of contours, but maybe something simpler based on triangulating 2 related contours will also work. It might be needed to make the relationship between subsequent contours explicit, e.g. by means of a contour tree data structure.

The second objective is to analyse the strengths and weaknesses of this approach, and possibly comparing the obtained surface to other methods to obtain such a surface (e.g. the approach of Zhou and Neumann [3], for which the source is available on github [4]).

Relevant research questions:

References:

Necessary skills : You will have to program in (at least) Python (and maybe C++) for this topic.

Contacts

Keywords

3D, Point clouds

Taking climate change into account in Land Administration: How to improve and evaluate existing systems 🔗

Description

This study aims to conduct a systematic review of land administration (LA) literature and international guidelines in the context of climate change to identify approaches and methodologies proposed on how LA functions are affected by climate change and how LA should respond to the challenges. It will also explore how current LA information management systems can be improved to support responsive land administration in the context of climate change. Considering the results of the research, an evaluation framework (or index) can be developed to assess climate change resilient land administration systems. The current study also aims to explore how the results of this research can be integrated into the Land Administration Domain Model (LADM).

Relevant literature:

Contacts

Keywords

Climate change, Database Management Systems, Land Administration

Temporal Analysis of Point Clouds for Identifying Significant Changes in Urban Landscapes (at Geodelta) 🔗

Description

This research project (graduation internship at Geodelta) aims to develop and explore methods for detecting significant differences between point clouds generated at different points in time. Specifically but not exclusively, the project will focus on comparing the AHN4 (Algemeen Hoogtemodel Nederland 4) and the subsequent AHN5, both of which represent the general height model of the Netherlands, captured a few years apart.

The objectives and challenges of the project include:

  1. Method Development: Creating robust algorithms and techniques to compare point clouds from different time periods that can handle the vast amount of data in point clouds.
  2. Change Detection and classification: Identify changes in the landscape, such as alterations in buildings, newly constructed structures, and other notable modifications and use this to develop criteria to determine what constitutes a major difference and create methods for accurately classifying these changes.
  3. Assessing the Reliability: Assessing the reliability of the methods for varying practical applications and at different scales in urban planning and development.

By accurately identifying and analyzing changes in point clouds over time, this project will provide valuable insights for municipalities in monitoring urban development, planning new infrastructure, and maintaining up-to-date geospatial data.

Temporal Analysis of Point Clouds

Geodelta is looking for a motivated student who is willing to make a deep-dive in point clouds and software engineering. While we are happy to provide a lot of support, the following competences will give you a head-start in your internship:

At Geodelta, we know that measuring is more than just precision. It is the reliability that counts. Are the measurements correct and have the correct procedures been applied? We are engineering consultants focussed on delivering the best technical advice and software in the fields of geodesy, photogrammetry and laser scanning. We make sure that our clients can trust their measurements. We do this by developing user-friendly software and by providing trusted advice. We are independent, innovative, pragmatic and above all we have a passion for geodesy. We are based in the century-old former faculty building for Geodesy. This is just around the corner of the building of the faculty of Architecture. If you are interested in this topic, or if you have an amazing idea yourself, let us know! We will arrange meetings with interested students discuss the project(s) and find the best fit.

Contacts

Keywords

3D, Change detection, Point clouds

Temporal point clouds to detect weathering of walls 🔗

Description

Various processes, including salt weathering, frost damage, sulfation, are all process which manifest themselves at the surface of the material of building walls and cause discoloration, soiling, loss of cohesion and/or adhesion, resulting in surface recession. Please find below some pictures showing example of damage to a brick and stone masonry due to weathering. In order to take timely action, it is omportant to monitor this type of decay in time. As mentioned, this damage takes from months to years to develop in the field, but it can be accelerated in laboratory.

church.jpg brik.jpg

Contacts

Keywords

Point clouds

The LADM as Linked Open Data 🔗

Description

The focus of this assignment is on designing a Linked Data profile for the Land Administration Model (LADM). The LADM is an ISO standard, and currently several profiles are being developed to become part of the new standard release. With this assignment we would like to transform the model into an RDF Ontology. Next to the ontology, instance data should be created, including a data story that can be used to show the value of having a Linked Data profile for LADM.

The assignment can be carried out at Kadaster, or in close collaboration with Kadaster. linked_data.png

What is Linked Data?

Linked Data is a method of publishing structured data so that it can be linked and queried. It is built on the Semantic Web technology, and is driven by open standards set by the World Wide Web Consortium (W3C ). The term was invented by Sir Tim Berners-Lee who also set the four design rules that are often used to define linked data. They are as follows:

  1. Use URIs as names for things
  2. Use HTTP URIs so that people can look up those names
  3. When someone looks up a URI, provide useful information, using the standards (RDF and SPARQL)
  4. Include links to other URIs so that people can discover more things

Although these design rules look simple, it is far more complicated when diving into the world of linked data. The first two principles touch upon the use of HTTP URIs to name things. In the concept of the Semantic Web HTTP URIs are used as names for real-world objects and abstract concepts rather than as addresses for Web documents. The content of a data set is structured using a simple graph-based data model-- the Resource Description Framework (RDF) . In RDF, a resource is described as a set of statements called triples or facts. A triple represents the basic structure of a sentence consisting of three parts, namely a subject, predicate and an object.

What is LADM?

The land administration domain model (LADM) provides a formal mechanism for describing land administration data, such as cadastral and deeds data. It is implemented by developing an application schema or profile of the LADM model.

It provides an extensible basis for the development and refinement of efficient and effective land administration systems, based on a Model Driven Architecture (MDA), and enables involved parties, both within one country and between different countries, to communicate, based on the shared vocabulary (that is, an ontology), implied by the model. The latter is required for the sharing and exchange of data.

Contacts

Keywords

Geoweb, Land Administration

Urban form, climate and personal comfort in a highly dense urban environment: The case of Seoul, South Korea 🔗

Description

The widespread availability of large datasets in cities is transforming the way urban systems are planned, simulated and visualized. However, data-driven initiatives often neglect the central role of urban dwellers, whose activities create the demand for energy and mobility in urban areas. In line with this, a better understanding of the relation between urban form, urban climate, and indoor and outdoor thermal comfort is important for an affective (re)design of urban systems. A potential thesis topic is the exploration of this relationship through the leverage of (existing) multidisciplinary datasets, comprising on the one hand high resolution datasets on urban climate, building energy performance, and surface temperatures, and on the other hand, thermal comfort, physiological and location data from participants in a study in Seoul, South Korea. The ultimate goal of this project is to demonstrate how open data combined with subjective feedback can help make informed decisions in planning outdoor spaces and providing for outdoor thermal comfort in a high-density urban area.

In this project, the student will explore the integration of these datasets to assess the urban features that most affected the participants’ perception in a highly dense urban environment. She/he will first create a model for the urban climate in the case study city, for example through data interpolation from high resolution weather stations and/or through the use of satellite imagery. Subsequently, she/he will work on exploring machine learning techniques to understand the correlation between participants’ subjective responses with features such as the urban climate, urban form features, and their physiology, activities and behaviors. This will enable the prediction of occupant’s outdoor thermal comfort based on the most relevant features in the urban dataset, as well as exploring how wayfinding and stay points in an urban district might be affected by the aforementioned factors.

Required Skill: Programming

Urban form

Contacts

Keywords

Artificial Intelligence, Thermal comfort, Urban form

Usability of the vario-scale approach in interactive and dynamic mapping 🔗

Description

Humans have been using the concept of map scale since the early days. When moving to an interactive and dynamic digital environment, the old map scale concept is maintained (even though the term "scale" may be replaced by "resolution") and for each different scale the whole map dataset is separately produced and stored. This is a labour intensive and expensive process and results in inconsistencies in practice. There are already map interfaces providing the feeling of vario-scale by supporting smooth zoom; e.g. Google Maps and/or Microsoft Bing Maps. However, this is just an illusion as "under the hood", the solution is still based on a number of redundant and fixed map scale representations.

We propose a new concept named 'vario-scale' in which the data once stored in a vario-scale data structure can be used for generating all wanted scales in a gradual and possibly smooth way. The vario-scale project ( http://varioscale.bk.tudelft.nl/ ) started out with the hypothesis that it is better to have small steps (content zoom) during a zoom operation. This then leads to map interaction where a delta in scale (small zoom step) leads to a small change in content on the map.

We are now looking for MSc students who are interested in carrying out a usability study of the vario-scale approach.

In this thesis research project vario-scale maps will be compared with fixed scale maps by designing and carrying out a usability study. The set up of the study entails eye tracking in combination with thinking aloud, screen logging, etc. when users carry out realistic map-use-tasks. For the comparison, the initialWebGL vario-scale viewer prototype ( https://varioscale.bk.tudelft.nl/gpudemo/2019/12/drenthe/ ) can be used (and possibly extended with additional functionality - e.g. other types of smooth zoom interaction).

The usability study will be carried out in the usability lab of the University of Twente (ITC), where eye tracking hardware, as well as software for analysis is available.

ssc+gpudemo.png

eyetracking.gif

Another prototype can be used to compare can be found at https://pengdlzn.github.io/webmaps/2020/11/interface/drenthe-comparer-swiper.html , where the left map uses a multi-scale layer as foregound and the right map uses a vario-scale layer as foreground. Both maps use a multi-scale raster layer as background. The following picture is a screenshot of the prototype.

20201204_drenthe-comparer-juxtaposition.png

Contacts

Keywords

Vario-scale

Using Foreign Data Wrappers inPostreSQL /PostGIS to expose large point clouds stored on the file system as LAZ files. 🔗

Description

It's an advantage to store LiDAR data into a Geographic Database Management System (Geo-DBMS) as this brings the possiblity of easily combining the LiDAR data with other types of geographic data, see e.g. [3]. In PostgreSQL a mechanism exists, called: Foreign Data Wrappers (FDW). "Foreign data wrappers (FDW for short) allow developers to expose external sources of data as tables inside PostgreSQL." [2] Oslandia has made an initial implementation for storing point clouds (but is not yet supporting .LAZ files) [4, 5]. The goal of this project is to make the .LAZ files from the AHN dataset available in PostreSQL/PostGIS by means of Foreign Data Wrappers and report on the benefits/problems encountered while doing so (compared to loading/copying the points into the database).

This work can extend the earlier work of Mutian Deng, e.g. by making an implementation using PGRX or https://supabase.github.io/wrappers/.

References

Contacts

Keywords

Point clouds

Using LiDAR to update a cadastral dataset 🔗

Description

The Queensland Cadastre consists of property boundaries defining the extents of around 2 million land parcels of differing sizes, from small urban lots to many thousands of square kilometres. The positional accuracy is similarly variable. For example, in a rural town, the dimensions of land parcels may be quite accurate, but the absolute positional accuracy of whole town may be significantly in error. The Department is obtaining high density LIDAR data, accompanied by high resolution photography, which may be used to improve the absolute positional accuracy of the Cadastral database (the DCDB), and is interested in the possibility of automatic reconciliation of these data. The aim would be to determine a set of displacement vectors that would place the corners of cadastral parcels in the most appropriate position in relation to the imagery, using identified fence lines and other such features for reference.

lidarproject.001.png

More information about the project: http://www.ergon.com.au/community--and--our-network/trees-and-powerlines/roames

Contacts

Keywords

Point clouds

Vario-scale Web Maps 🔗

Description

Humans have been using the concept of map scale since the early days. When moving to a digital environment, the old map scale concept is maintained and for each different scale the whole data of a map are separately produced and stored. This is a labor intensive and expensive process and in practice results in inconsistencies. There are already map user interfaces providing the feeling of vario-scale by supporting smooth zoom; e.g. Google Maps and/or Microsoft Bing Maps. However, this is just an illusion as the solution behind the "curtains" is still based on a number of redundant and fixed map scale representations. We propose a new concept named "vario-scale" where the data once stored in a vario-scale data structure can be used for generating all wanted scales in a gradual and possibly smooth way. We look for Msc students that are interested in solving topics related to automated map generalization, new ways of visualization, data management and usability testing for this new type of digital maps.

Website : http://varioscale.bk.tudelft.nl Requirements and skills : Programming

Contacts

Keywords

Geoweb, Vario-scale

Visualisation of different types of 3D spatial units 🔗

Description

Dense urbanisation has led to an increasing demand and pressure for land development, resulting in the partition of 3D space into different owners sharing delimited property interests on, above or below the land surface. Consequently, (cadastral) spatial units range from simple, but most common, 2D, to complex 3D collections of spaces worldwide, that are more difficult to handle in terms of surveying, storing in a database, maintaining, visualising, etc. The complexity of volumetric spatial units worldwide is highly variable, namely: condominiums, archaeological sites, mines, pipelines, bridges, marine parcels, air parcels, etc. In order to be used for visualization, the various types for spatial units, should be stored in a format appropriate for downstream use.

An MSc thesis topic could focus on the visualisation of different types of 3D spatial units in a web-based 3D cadastral prototype (previous research can be used as basis; namely: Web-based visualization of 3D cadastre). In this context, pertinent standards in 3D storage and visualization relate to both data format and grammar, implementation as with programming interfaces (API) and Web Feature Services should be explored (many of them are proposed by ISO, OGC andW3C ). Use cases from various types of volumetric spatial units can be used to test the prototype that will be developed.

3D_Cad3.jpg

Related literature :

Contacts

Keywords

Land Administration

Visualizing and disseminating property valuation information 🔗

Description

Property valuation is a process of forming an opinion of value of immovable property under certain assumptions. It is performed by public and private sector actors for several LA processes, such as property taxation, compensation on expropriation, land readjustment, land consolidation, insurance assessment, real estate financing, and property transactions. Appropriate systems are required for effectively managing, visualizing and disseminating property valuation related datasets. The Valuation Information Package of ISO 19152 Land Administration Domain Model (LADM) provides a common basis to develop holistic prototype systems for property valuation information.

Objective: The aim of this study is to design and develop an LADM compliant 3D prototype system for visualizing and disseminating property valuation information and creating interactive 3D property valuation maps for the Netherlands.

ladm_valuation1.jpg ladm_valuation2.png

Relevant literature:

Contacts

Keywords

Land Administration