IBM Research Zürich is working on an inspection project together with Canton of Zurich at Dübendorf Air Base. The focus here is on the development of a new class of foundation models for visual inspections on the basis of Artificial Intelligence (AI). In this way, the aim is to detect cracks in civil infrastructures such as bridges, roads and airport runways, as well as improving maintenance procedures.
The foundation model is designed in such a way that it can learn a general representation of concrete surfaces and runways. Once it has been fine-tuned to the specific conditions of the runway in Dübendorf, for example, it can be used to detect cracks in the surface of the air base’s runway. According to a press release issued by IBM, the runway at Dübendorf will be visually scanned by a drone fitted with a camera from the Bern-based company Pixmap. The AI model will then be tasked with identifying cracks in more than 10,000 images using a method known as instance segmentation.
The Dübendorf Inspection Project follows on from previous projects of this kind. In Denmark, for example, Europe’s longest suspension bridge, the Storebælt, was examined for the most minute cracks, spalling, algae and rust. At Frankfurt Airport, the task was to identify obstacles on runways such as cans, bottles, garbage and small pieces of metal. IBM Research is firm in its conviction that the future of AI lies in such flexible, reusable AI models that can be applied to almost any industrial area or task.
