AMAzE 2.0 – Digital building application submission with a focus on greening of buildings and land, and the degree of soil sealing

AMAzE 2.0 developed a prototype workflow for digital planning applications and subsequent monitoring, with a focus on green spaces and land sealing. By integrating GIS, BIM and AI-supported remote sensing analyses, heterogeneous data was converted into consistent digital models, thereby creating a basis for automated review processes. The project thus supports efficient planning approval procedures and sustainable urban development.

Short Description

The increasing urbanization as well as rising demands for sustainable and climate-resilient urban development require new data-driven approaches for the planning, approval, and monitoring of construction projects. In particular, the consideration of green spaces and surface sealing plays a central role in the context of climate change adaptation, biodiversity, and urban quality of life. At the same time, existing building permit processes are often characterized by heterogeneous data formats and manual verification procedures, resulting in inefficiencies and limited traceability. Against this background, the AMAzE 2.0 project addresses the research question of the extent to which an end-to-end digital process chain based on Building Information Modeling (BIM), Geographic Information Systems (GIS), and methods of Artificial Intelligence (AI) can be developed to enable an automated and verifiable building permit process with a particular focus on the outdoor environment.

The current state of the art is characterized by fragmented system landscapes and insufficient interoperability between GIS and BIM systems. While BIM is increasingly established as a standard for digital building modeling, essential planning data such as cadastral information, terrain models, or environmental data are predominantly available in GIS-based formats. Seamless integration of these data is currently only possible to a limited extent. In addition, there are deficits in the automated verification of submissions, particularly in the area of outdoor spaces. At the same time, modern remote sensing methods offer significant potential for objective and automated as-built analysis but are rarely systematically integrated into approval processes. A further major limitation is the restricted availability of standardized interfaces.

The objective of the AMAzE 2.0 project was the development and prototypical implementation of an end-to-end digital workflow for building permit submission as well as the subsequent monitoring of construction projects, with a particular focus on outdoor spaces. The focus was on the integration of GIS and BIM data within a consistent data pipeline, the automated transformation of heterogeneous input data into standardized IFC (Industry Foundation Classes) models, and the definition of information requirements for BIM-based submission models. In addition, AI-based methods for the analysis of aerial imagery and point cloud data were developed, and an as-planned versus as-built comparison was prototypically implemented.

The methodological approach is based on the integrated combination of GIS, BIM, ETL-based data processing (Extract, Transform, Load), and AI. At the core of this approach is an ETL-based data pipeline implemented using FME (Feature Manipulation Engine), which integrates heterogeneous geospatial data and transforms them into consistent, semantically structured IFC models. Complementary script-based processes allow for precise control of the model structure and ensure the quality and stability of the generated BIM data. In parallel, a modular AI pipeline was developed, covering data extraction, semantic segmentation, as well as statistical and visual evaluation. The use of spectral indices such as Normalized Difference Vegetation Index (NDVI) further improves the quality of green space detection.

Another key contribution of the project lies in the development of AI-based methods for the automated evaluation of aerial imagery and point cloud data. The applied neural networks enable pixel-accurate segmentation and provide robust classification results for different surface classes. The evaluation shows that the combination of remote sensing data and AI-based analysis provides a suitable basis for automated as-planned versus as-built comparison, although challenges remain regarding data availability and system interfaces.

Based on the achieved results, several directions for future development emerge. A central focus lies on the transition towards a fully API-based system architecture (Application Programming Interface) in order to reduce existing media discontinuities. Furthermore, there is potential for improving AI models through larger and more diverse training datasets as well as through the integration of additional data sources. Another important step is the transfer of the prototypical solutions into operational applications within administrative environments. This requires, in particular, the establishment of appropriate legal and organizational frameworks. In addition, stronger standardization of data models and interfaces is necessary to ensure long-term interoperability. Overall, the approach developed within the project can make a significant contribution to the digitalization and automation of building permit processes and thus support more efficient, transparent, and sustainable urban development.

AMAzE 2.0 – Digital building application submission with a focus on greening of buildings and land, and the degree of soil sealing

Short Description

Contact Address

Projektleitung

Projektpartner