Former Projects
Cool Street Project
Total Duration: 2 M.Sc. for 1 year
The CoolStreet project collaborating with ClimateFlux GmbH. aims to develop a service platform in the format of an API to generate a map-based application that displays the temporal-spatial distribution of urban heat to allows pedestrians and cyclists to choose alternative routes to minimize their exposure to heat stress. Unlike conventional routing applications mainly focusing on travel distance and time, this project addresses the environmental influences on pedestrians and cyclists, promotes thermal comfort routes and aims to improve the awareness of city climate. In this project, we will adopt predictive computer modeling methods (AI) and data-driven approaches to predict micro-scale environmental exposure levels. The ML algorithm is trained on a large amount of building and tree height models to generate hourly shadow distribution and an integrated urban comfort index will be formed and applied as one of the weighting attributes of road segment to identify healthier routes in road network of Munich city. A user-centered interface will be designed to support the customized navigation services.
Dense and Deep Geographic Virtual Knowledge Graphs for Visual Analysis
Total Duration: 1 doctoral candidate for 3 years
The DFG project has the goal to effectively integrate and analyze heterogeneous geodata sources by bridging two research fields – Virtual Knowledge Graphs (VKG) and Geovisual Analytics. In the project we will develop on the one hand a set of methodologies and software tools to create densely and deeply linked geospatial knowledge graphs, and on the other hand a set of intuitive visualizations and explainable analytical services over them. The project will provide a complete solution covering the whole life-cycle from the construction and enrichment of GeoVKGs, to their visualization and analysis over them. The developed methodology and software tools will be demonstrated in two real-world use cases. The first one demonstrates climate change in Bavaria, while the second one deals with tourism and mobility data in South Tyrol, managed by NOI Techpark in Bolzano. Although the two use cases have different characteristics, they both represent common and intricate cases of geodata management, and thus require the ability to effectively combine the underlying datasets and to perform complex analytical tasks over them.
OpenStreetMap Boosting using Simulation-Based Remote Sensing Data Fusion - OSMSim
Total Duration: 2 doctoral candidates for 3 years
This DFG project aims to improve building information in OpenStreetMap (OSM) using a simulation-based fusion of heterogeneous remote sensing data and the updated OSM data for follow-up applications. The simulation environment SimGeoI provides the starting point. It is used to compare geometric OSM information with remote sensing data produced under different sensor configurations and at different acquisition times, and to enrich OSM with geometric corrections (position, height) and attributes (e.g. building type, roof structure) gained from a fusion of the different remote sensing data. Three core themes are addressed: First, a methodical framework will be developed, which allows the geometric correction of OpenStreetMap data based on the prediction and comparison of building shapes, using a pair of remote sensing images (optical, SAR or mixed). In a second step, geometrically improved OSM information will be used to extract building-related attributes from multi-modal remote sensing data. Finally, the transferability of the developed methods will be experimentally analyzed and interfaces to follow-up applications will be investigated. The methodology will be accompanied by validation in order to evaluate the positional, thematic and temporal accuracy of derived results.
Guided Unlearning of Cognitive Pitfalls in Georeferenced Social Sensing
Total Duration: 1 postdoc researcher or 1 doctoral candidate for three years
This DFG project addresses bias-induced cognitive pitfalls in social sensing. For selected application scenarios, an interactive platform for guided unlearning of cognitive biases will be developed and prototypically implemented. Unlearning is a radical method of learning. Unlike conventional learning or knowledge accumulation, which is based on the addition of what is new to the learner, an unlearning process is based on the conscious subtraction of something undesirable that already exists, either innately or learned. The project has three objectives: to improve the transparency with regard to the value chain of georeferenced social sensing; to support users’ holistic understanding of cognitive biases in georeferenced social sensing; and to assess users’ capacity of critical reasoning after the training of guided unlearning. A number of guided unlearning experiments will be designed and implemented with biased data for selected real-world scenarios. We will collect preliminary findings and/or raise new questions, relying on two kinds of comparison: between untrained user solutions and benchmark solutions, and between untrained and trained user solutions.
Spatial association and GeoAI
Begin: since 2020.10
Contact: Peng Luo
This project aims to deepen our understanding of spatial association and develop the new Geospatial Artificial Intelligence model (GeoAI). First, this project explores a new way to understand spatial association apart from spatial heterogeneity and spatial autocorrelation. Second, this project attempt to introduce a new understanding of spatial association into AI algorithms. The developed new GeoAI models will be applied to urban computing, social sensing, spatial optimization, population mapping, and other fields.
Data-driven exploration and explanation of ethics cases in business world
Begin: since 2021.09
Contact: Chuan Chen / Mengyi Wei
This project funded by industry partner addresses the ethical issues in business world from a data scientific perspective. The ethics cases in big data are mainly reflected in narratives of conflicts, negotiation contents, trial processes and results, professional comments, public opinions, analytical judgments, and constantly updated lists of "should do" and "should not do". However, without AI support, the human brain alone cannot see through the intricate and varying strengths and weaknesses of connections among the elements of a case or among different cases. In this project we attempt to develop an interactive visual analytical platform that can combine the data-driven deep learning ability with knowledge-driven human's reasoning and interpretation ability. The platform will enable us to perform tasks along a value chain such as automatic collection of ethics cases from the globally accessible public media and social media, analysis and clustering of collected data, creation of a specific ontology and knowledge graph, and visual explanation of the knowledge graph. Being driven by big data, the platform may serve the general objectives for different stakeholders in the business world.
Situated Geovisualization Based on Mixed Reality
Begin: since 2020.12
Contact: Shengkai Wang
Limited spatial information obtained from the real world leads to poor human spatial memory and navigation performance. Mixed reality (MR) enriches accessible information and improves the interaction experience of users' surroundings, reflecting great potential in improving human spatial cognition in recognizing, understanding, memorizing, retrieving, and representing. However, factors like spatial scale and information complexity may have impacts on the usability of MR-based visualizations and thus cause distortion and loss of spatial memory. This project addresses the human spatial cognition issues from the perspective of visualization based on MR. Our research is to explore the behaviors of users, uncover the impacts of visual and spatial factors on usability, and develop an MR-based visualization and visual analysis platform for improving human spatial cognition.
Semantic Data Mining for OSM Building Layer
Duration: 2022 - 2023
Contact: Leul Kahsay/ Peng Luo
One of the fundamental supporting technologies of AR map is to build a complete, accurate, consistent and up-to-date 3D city model dataset. The future success of the AR map application relies on the capability of extending its coverage to city-level and even larger areas. For the moment, the major data sources for the ultimate experience areas include LiDAR point clouds, aerial imagery and panoramic images. The data collection is costly and requires large amount of manual work. While for the city-level 3D model construction, one of the approaches is the use of stereo satellite imagery. This method is able to provide the global coverage, but has a relatively long repetition cycle (more than three years) and is severely affected by weather conditions. Besides, its cost per square kilometer is also higher than for normal satellite imagery because it has to be in stereo. This project funded by industry partner addresses the scalability of single-scene SAR imagery and VGI building footprints; and the comprehensibility of historical building attribute data and land use data for the creation of 3D city models.
Mixed Reality-based Indoor Navigation and Spatial Learning
Duration: 2019 - 2022
Contact: Bing Liu
Modern people spend most of their time indoors, and they move a lot within closed spaces. Indoor navigation is an integral part of our life. The indoor navigation applications are much more limited compared with outdoor. The main reason is the difficulty of getting stable GNSS signals. The quickly developing mixed reality (MR) technology performs well in indoor spatial mapping and indoor positioning. Head-mounted MR is highly potential in indoor navigation, there are risks that the wrong virtual information misleads the users or the perception of the physical world is decreased. The users’ perception and usage preferences would improve usability and accelerate the maturing of MR-based indoor navigation. In this ongoing project funded by China Scholarship Council, we collect the ordinary users’ attitude toward using head-mounted MR for navigation, explore how MR-based navigation influence spatial learning and find ways to improve its usability.
Map-based Dashboard
Duration: 2018 - 2022
Contact: Chenyu Zuo
Map-based dashboards are among the most popular tools that support the viewing and understanding of a large amount of geo-data with complex relations. In spite of many existing design examples, little is known about their impacts on users and whether they match the information demand and expectations of target users. We first designed a novel map-based dashboard to support their target users' spatiotemporal knowledge acquisition and analysis, and then conducted an experiment to assess the feasibility of the proposed dashboard. The experiment consists of eye-tracking, benchmark tasks, and interviews. Forty participants were recruited for the experiment. The results have verified the effectiveness and efficiency of the proposed map-based dashboard in supporting the given tasks. At the same time, the experiment has revealed a number of aspects for improvement related to the layout design, the labeling of multiple panels and the integration of visual analytical elements in map-based dashboards, as well as future user studies.
Semantically Enriched and User Orientated Multi-Modal Navigation
Duration: 2016 - 2020
Contact: Christian Murphy, Edyta Bogucka, Linfang Ding
The Federal Ministry of Transport and Digital Infrastructure (BMVI) leads a data-oriented R&D- funding programme for the period 2016 to 2020 in a form of the modernity fund (mFUND). Within this programme, Chair of Cartography TUM conducts a research on early developments of digital innovations in mobility. Ongoing project aims to show the added value of extending the existing traffic data with semantic information. The pre-study examines the possibilities of (1) the enrichment of the multimodal traffic database with semantic information to be gained from user behaviour data, Volunteered Geographic Information (VGI) and social media and (2) development of the user-oriented multimodal navigation service. As a result, a demo app version will be developed to present the following scenarios relevant for the urban and mobility design: multiscale representation of transportation nodes, automatic detection of the smombie danger and managing hotspots of negative events in the cities. Semantically enriched, multimodal and user-oriented navigation service will be evaluated in two test areas of Berlin and Munich.
Geospatial Information Services for Smart Cities Driven by Big Data
Durtaion: 2017 - 2020
Contact: Edyta Bogucka, Linfang Ding
Partners: Yangtze River Delta Science Data Center, Changshu Fengfan Power Equipment Co., Ltd
Within this cooperation, Chair of Cartography TUM conducts a research on integrating multiple sources of geo-referenced urban data to derive valuable geospatial services for smart city applications. As a result, an open geo-collaborative portal with a set of interactive services will be developed. The portal will contain an interactive user interface, an extendable visualization toolkit and an analytical toolkit. The anticipated components will serve as descriptive, diagnostic, predictive and prescriptive tools for city management. The functionalities of the portal will be developed during the series of workshops and user tests with three target groups – urban citizens, domain experts and decision makers. Each target group will participate in the knowledge construction process in a two-fold role – as information receiver and data contributor.
A Visual Computing Platform for the Industrial Innovation Environment in Yangtze River Delta
Duration: 2018 - 2020
Contact: Chenyu Zuo, Linfang Ding
Partners: Jiangsu Industrial Technology Research Institute (JITRI), Yangtze River Delta Science Data Center, School of Geography - Nanjing Normal University
The aim of the joint project is to develop a visual computing platform dedicated to monitoring the dynamic innovation and investment ecosystem in Yangtze River Delta. The platform combines the power of intuitive human vision with that of analytical computing, thus serves as an enabler for users to explore the interactions between the regularly updated geo-infrastructure data and the continuously evolving geo-economic data, and to preview the complex influence factors of industrial innovation. The anticipated platform will be prototypically implemented with three extendable components – a geovisualization toolkit, a geospatial analytical computing toolkit and a geo-economic event collector. These components will be tailored to the needs and knowledge profiles of three target groups who are involved or interested in the innovation ecosystem in Yangtze River Delta – governmental agencies, industrial enterprises and investors.
Sense-Making Image Mapping from Remotely Sensed GLC30m
Contact: Ekaterina, Chuprikova
This joint project between the Lehrstuhl für Kartographie at TUM (TUM-LfK) and the National Geomatics Centre of China (NGCC) aims to visually empower the uniquely available GLC30m (global land cover of 30m resolution) at NGCC with the uniquely available attentionguiding design framework of concise image maps at TUM-LfK. Both partners are committed to developing a globally accessible open-source platform for GLC30m. The platform will provide a metadata catalogue and (semi)automatic value-adding services to enable the continuous validation, updating and efficient use of GLC30m incl. visual query, visual narratives of query results and creation of web-based image maps for selected applications and target groups. Across the fields of remote sensing, geoinformatics, visual perception and neuropsychology and cartography, innovative methods are anticipated to (a) interlace the first-hand information from raster images with the second-hand information from map symbols and labels at multiple visual levels rather than just a figure-ground composition; (b) unite the pixel resolution reflecting the degree of ground-truth with the map scale reflecting the cognitive abstraction in the visual storytelling; (c) convert the globally accessible land cover types into personalized visualizations for efficient data understanding.
Event detection and visualization of Volunteered Geographic Information (VGI)
Contact: Polous, Khatereh
With rapid spread concept that uses web as “participatory platform”, assessing spatio-temporal processes and detailed change mapping, which highly demand accurate and up-to-date data, has become more affordable. Many studies have been already conducted for detection, monitoring and visualization of changes from multi-temporal, multi-spectral and multi-sensor data. But, it is less discussed how the detected changes should be decomposed and formulated to reveal an event. The aim of this study is to detect location-based events from Volunteered Geographic Information (VGI) in Munich. The work concentrates on detection and pattern recognition of events, which are bound to a specific time and place from delivered information by internet users.
Analysis and Conflation of Road Networks in Digital Maps for Automotive Applications
Contact: Andreas, Hacklöer
A joint effort with BMW Research and Technology, this project investigates methods for the analysis and conflation of road networks. Modern car navigation services rely on vector-oriented road models embedded in digital maps. In order to identify a given road network structure across multiple maps, a matching and conflation process is required which determines a projection from one road network to another, thereby providing a map-agnostic means of identifying geographical references to entities describing road networks. In the project, we develop, investigate and evaluate road network matching techniques for digital vector maps which analyze geometrical, topological and semantic information to derive assignments between road networks on several abstraction levels, ranging from point-to-point correspondencies to structural high-level matchings. The resulting algorithms represent an enabling technology for comparison, fusion, attribute transfer and quality evaluation of digital maps.
Spatial Optimization of Railway Infrastructure Maintenance for Travel Time Saving
Contact: Jian Yang
In this ongoing project, we study the problem by how can we reduce minute-level running time of passenger trains on a given railway route through maintenance instead of costly construction work like changing the existing railway alignment. This interdisciplinary project leverages the expertise not only from cartography but also from railway infrastructure construction, transportation planning and computer science. So far, we’ve established a five-step semi-automated process including data preprocessing, timesaving potential estimation, maintenance site selection, site inspection and planning result reporting and also developed a prototype program tool based on our method to facilitate decision makers on maintenance planning.
Comparative study of thematic mapping and scientific visualization
Contact: Linfang, Ding
Visualization of 4-D Building Deformation from High-resolution SAR Data — The project is dedicated to visualize 4-D space-time building deformation datasets retrieved from high resolution satellite SAR images, which are crucial to monitor building behavior as well as detect potential damages in urban areas. Based on the data characteristics, appropriate design methods and visualization techniques should be identified from disciplines like scientific visualization and thematic mapping, which may allow users to explore detailed deformation information as well as to perceive the overall building deformation patterns immediately.
Lightning data analysis
Contact: Stefan, Peters
This research project focus on the visual analysis of lightning data. Starting point are 3D coordinates and the exact occurrence time of lightning data. In a first step lightning cells are identified and tracked. Then an interactive graphic user interface is developed to investigate the dynamics of the lightning cells: e.g. changes of cell density, location, extension as well as merging and splitting in 3D over time. Furthermore a statistical analysis is provided. The second part of this research contains a short term forecasting of lightning cells and the visualization of its uncertainty. The visual exploring tools are investigated for two determined user groups: lightning experts and interested lay public.
Multi-dimensional visualization of spatio-temporal data
Contact: Christian Murphy
Space-Time Mapping of Mass Event Data — To simultaneously visualize thematic data in space and time a third dimension must be added. In this work conventional cartographic symbolization meets the space-time cube to create a holistic three dimensional spatio-temporal visualisation model. The two dimensional proportional symbol mapping technique is adopted and extruded into the third dimension to model the temporal factor. Kernel density estimation is performed on the time line to create a temporal continuous model from discrete points in time. The resulting visualization model is implemented into an earth viewer to enable the user to freely navigate and animate the phenomenon and visually detect spatio-temporal anomalies without losing the overall view. This tool is evaluated by visualizing the events of a mobile phone location dataset over space and time in one single model.
Multi-dimensional visualization of spatio-temporal data
Contact: Mathias Jahnke
To support experts as well as non-experts in their decisions is one main goal of the geospatial domain. The non-photorealistic visualization offers a new way of a user oriented cartographic communication on small mobile displays. In particular 3D city models presentations can profit from this approach because most of the city model visualizations do not tap the full potential of a combined geometric, semantic information presentation. The non-photorealistic approach of abstracted information reduced visualization seems to be feasible for a combination with cartographic design concepts and offers more degrees of freedom to add semantic information particularly on small mobile devices. To reach this goal on one hand the theories and method from traditional 2D cartography have to be expanded for the usage in the third dimension on the other hand the user have to be taken into account. The user is of main interest when developing new visualization styles because he is the only person who can give valid feedback about the usability of such new visualization styles.
2D and 3D Thermal facade data visualization
Contact: Holger Kumke
The project is dedicated to the visual enrichment of thermal data on building facades in urban area. Heat radiation or thermogram which can be detected and measured by physical instruments shows invisible thermal information and indicates the state of the building surface. It serves as raw data for the design of thermal façade maps on planar display surface as well as 3D space-based map-related depictions. Similar to the digital thematic map design, the thermal maps are conceived for the output on screens. Their special characteristics, however, have opened up many innovative visualization alternatives.
Location-Based Services & indoor navigation
Contact person: Jukka Krisp
Location-Based Services (LBS) are investigated from different perspectives that include mobile positioning and tracking technologies, data capturing and computing devices, integrated software engineering, user studies for various applications. The enabling sensory technologies and interactive open-source platforms are continuously changing the way of our thinking and living and reshaping the research scope of LBS. LBS is an integrative discipline that unites research ideas from related fields. Technical challenges still occur on the indoor navigation data acquisition, the path computation and the communication of a potential indoor path to the user.
A Congruent Hybrid Model for Conflation of Geo-Referenced Image and Road Network
Contact person: Jiantong Zhang
In the project, we investigate a novel Congruent Hybrid Model(CHM) to rectify the misalignments between road vectors and geo-referenced images. The matching cost between the extracted road centerlines and the prior road network was optimized using Sparse Matching Algorithm (SMA) to get the optimal correspondence, and then the road segments were transformed to its partner using two frequently transformation functions — the piecewise Rubber-Sheeting (RUB) approach and the Thin Plate Splines (TPS) approach. The experiments with synthesized data as well as the real spatial data sets have verified the efficiency of CHM. The snake-based approach is a natural subsequence of the presented model. However, the CHM can be also directly employed for geospatial visualization applications
A climate event portal for knowledge discovery
Begin: since 2018.5
Contact: Liqiu Meng, Andreas Divanis
Partners: Bayerische Klimaforschungsnetzwerk (BayKlif): Prof. Dr. Annette Menzel, Technische Universität München, Prof. Dr. Dieter Kranzlmüller, Leibniz-Rechenzentrum, Prof. Dr. Susanne Jochner-Oette, Katholische Universität Eichstätt-Ingolstadt, Prof. Dr. Jörg Ewald, Hochschule Weihenstephan-Triesdorf, Prof. Dr. Wolfgang W. Weisser, Technische Universität München, Prof. Dr. Ulrike Ohl, Universität Augsburg, Prof. Dr. Arne Dittmer, Universität Regensburg, Prof. Dr. Henrike Rau, Ludwig-Maximilians-Universität München
The PhD or post doc researcher will be working with multidisciplinary teams in a Research Cluster on “Bavarian Synthesis Information Citizen Science Portal for Climate research and Scientific Communication” (BAYSICS - Bayerisches Synthese-Informations-Citizen Science Portal für Klimaforschung und Wissenschaftskommunikation).
Project webpage: https://www.bayklif.de/verbundprojekte/baysics/teilprojekt-3/