Landsight – smart data for community planning

Several Swedish municipalities, government agencies, and companies have recently highlighted a growing need to improve the analyses and decision-making tools used in urban planning and sustainable development. This encompasses everything from assessing stormwater flows and noise levels to simulating the impacts of new construction. At the same time, interest in digital twins—virtual replicas of cities used for communication, simulation, and visualization—has rapidly increased.

A key enabler of this shift is that beginning in 2025, within the framework of the EU initiative “High Value Data,” the Swedish mapping agency (Lantmäteriet) will make national aerial imagery and elevation data more accessible. These datasets have a much higher resolution (15 cm) than those currently used by municipalities — such as the Swedish Environmental Protection Agency’s national land-cover data (10x10 m) or commercial satellite imagery commonly employed on a global scale. Lantmäteriet’s data provides a much more accurate view of local features such as buildings and terrain, thereby offering the potential for more precise analyses.

In this project, Dubblett—a young company rooted in research at Linköping University—aims to capitalize on this new opportunity by systematically combining Lantmäteriet’s open aerial imagery with advanced AI based image analysis and 3D reconstruction. The objective is to build a web-based AI platform that automatically generates maps and 3D models tailored to Swedish conditions. By focusing on higher-resolution data compared to traditional satellite-based mapping, it becomes possible to produce more detailed classifications of vegetation, roads, waterways, and other infrastructure.

A critical element of the project is developing AI models that work across all of Sweden, whether in major cities or rural municipalities. Achieving this requires a sufficiently large volume of high-quality training data, which is both costly and time-consuming to label manually. Therefore, Dubblett plans to use synthetic data— artificial, computer-generated images that resemble real environments but can be adapted to cover a wide range of variations. Experience from Linköping University, particularly in medical technology, indicates that synthetic data can make AI models more robust and generalizable because they “see” more examples than one would encounter in a limited set of real images.

The project builds on several previous initiatives where image analysis and AI for urban development have already been tested on a smaller scale. Examples include the Visual Sweden City platform and Simstad two
research projects that produced prototypes for automated 3D modeling of buildings and natural features.

Further knowledge-building efforts such as Visual City have demonstrated municipalities’ desire for more standardized access to national geospatial data, and how digital twins can enhance transparency in the planning process. Additional projects within Smart Built Environment—such as 3CIM and Digital tvilling och analysverktyg — have shown that there is currently no comprehensive market solution offering both AI-based automation and high-resolution aerial data for the entire country.

To ensure that the new platform truly aligns with municipalities’ practical needs, five Swedish municipalities will form a reference group. They will provide continuous feedback on requirements, user interface, and data workflows. Areas of focus include simplifying stormwater and flood analysis, enabling noise and wind visualization, and offering evidence-based planning data for various development projects. Smaller municipalities, which often lack in-house GIS expertise, will thus gain access to the same advanced technology already available to larger cities and regions—albeit in a scalable and cost-effective manner.

Ultimately, the project aims to lay the foundation for a national solution in which AI and digital twins become integral tools in urban planning. Beyond municipalities, insurance companies, construction firms, and consultancies could benefit from data on aspects such as flood risks or green spaces. With its roots in academic research and extensive expertise in AI, synthetic data, and urban planning, Dubblett seeks to create a platform that facilitates more sustainable and efficient construction—thereby contributing to the future of vibrant and resilient urban environments across Sweden.