16.03, 16:30–16:50 (Europe/Berlin), Hörsaal 3 (0119)
Interior building data is gaining popularity, but while there is an abundance of exterior data for navigation and other purposes, the availability of indoor data is meagre. This paper presents how simplified indoor data can be extracted from digital building models, which are rich in interior information, and can be converted to formats such as CityGML, IndoorGML, and OpenstreetMap to increase the availability of indoor data.
Accessibility, barriers, and the locations of facilities inside buildings are gaining greater public interest. Accordingly, there is a popular trend to extend geospatial databases, map and routing services with indoor information, such that data and services can cater to broader applications. Various researchers address the topic by either extending geospatial data formats or by improving the quality and efficiency of indoor data acquisition. These works resort to existing 2D plans such as owner's as-built plans or publicly displayed escape plans as sources, or they create 3D models from laser scans or photographic material. However, these methods require substantial manual effort. Our work automates indoor data generation by leveraging existing data from construction planning, where building information is produced in good quantity and quality.
We develop conversion procedures that ingest building data as source, extract 2.5D floor plans, map them to an intermediate model and finally derive data in three pertinent target formats. We work systematically from the source and target models towards the intermediate model and take a bottom-up, inductive approach: On the source side, we analyse the IFC schema and realistic sample data for relevant patterns, implement checking routines and extract the input for conversion to the intermediate model. On the target side, we produce CityGML, IndoorGML and OSM sample data for a tailored sample project, identify commonalities of the schemas, design an intermediate model and generalize the output methods.
In the talk and paper, we present the identified IFC elements as well as the intermediate model in detail. Preliminary results indicate that the required source information exists in different forms among the semantic elements, geometric representations, and topological relations found in IFC. For the target models we realized that CityGML and IndoorGML are free of redundancies since they are both OGC schemas. Thus we identify a relevant subset of their union. OpenStreetMap is substantially different from the OGC formats. It is geometry-driven with semantic information attached to it. We use the Opensource BIMserver for efficient access to the often voluminous building data and implement the conversion in Java. A dedicated web application will serve as a front-end. Two university campus buildings as well as a public administration centre under construction are used for verification.
In summary, our work integrates the interior of publicly accessible buildings into the urban outdoor space, expanding the already available city models and databases. We focus on open formats, the production of open data, and publish our applications as open source. Building owners are given the opportunity to submit their data for public access, assisting in the process of volunteer mapping. The integrated data has the potential to provide the foundation for innovative applications in the area of autonomous navigation, accessibility and public transportation.
This paper is a substantially extended version of the lightening talk we presented at SOTM 2022. For more information, please visit https://bauinformatik.github.io/levelout.
Helga Tauscher studierte Architektur an der HTW Dresden sowie der Kunsthochschule Berlin-Weißensee und promovierte an der TU Dresden im Fachgebiet Bauinformatik. Sie war u.a. als CAD/CAFM-Fachfrau in der Aufzugsbranche und als Software-Entwicklerin für Internet-basiertes Bauprojektmanagement tätig sowie als Forscherin und Dozentin in Kaiserslautern und Singpore. Ihre Forschungsinteressen drehen sich um graphbasierte Modellintegration von Daten aus der Architektur und Bauwesen sowie um Opensource-Software-Entwicklung in der Branche. In 2022/23 hat sie eine Vertretungsprofessur an der Bauhaus-Universität Weimar inne.
Subhashini Krishnakumar studierte Geodäsie und Geoinformationswissenschaften an der TU Berlin. Wissenschaftlicher Mitarbeiter an der Bauhaus Universität Weimar im Projekt „Level Out“ seit 2022