By
Markus Lukasson
Share this post
Our research project ICAR has been approved! 🎉
ICAR stands for Image-to-CAD Assembly Retrieval. The project solves a persistent problem in industrial maintenance: identifying installed spare parts. Many components are dirty, hidden, or only visible after partial disassembly. A single photo is often not enough for exact identification. The AI has to understand the machine or plant as a whole. This is where we come in.
The real challenge, though, is scale. A single demonstrator is already feasible today. Reliable identification across hundreds of machines is a research problem. Every machine often exists in multiple revisions. Some are configured differently for every customer. The CAD data comes from different systems and varies widely in quality. Solving this is what we are working on.
Our goal: A part visible in a real photo is matched directly to its 3D counterpart in the CAD assembly. Service technicians then see the part in the context of the full machine. They can isolate it, view it in its installed position, and inspect subcomponents. And all of this robustly, across entire machine fleets.
ICAR builds on ten years of experience in visual spare parts search, used by customers such as Trumpf, DMG Mori, and Windmöller & Hölscher. The funding secures our research until October 2027. The results flow directly into our platform.
Thank you to Investitionsbank Berlin (IBB) for the trust and the smooth collaboration! A warm thank you as well to Berliner Strategen GmbH for the clear and well-structured guidance throughout the application process.
The research is co-funded by the European Union. The project is financed by the European Regional Development Fund (ERDF) and the State of Berlin. The grant is provided by Investitionsbank Berlin (IBB) through the ProFIT program.
