itsowl-TP-DeepConcrete: Framework for mapping prefabricated reinforced concrete parts from aerial images for intelligent outdoor storage optimisation

BREMER AG, based in Paderborn, is a German construction company that accompanies solutions in industrial and commercial construction - from the initial building idea through all planning phases to the construction and operation of a building. BREMER plans and produces all the prefabricated reinforced concrete components required for building construction in its own plants - predominantly in Paderborn.

As a long-term vision, BREMER AG is pursuing the complete digitalisation and intelligent automation of its order and goods flow management. The aim is to build a digital platform to automate and intelligently design the exchange between suppliers, customers and BREMER AG. The intended benefits are, for example, paperless construction planning, optimised logistics and intelligent production planning.

As a basis for an automated storage system, an automated, continuous recording of the actual storage status of BREMER AG's external warehouse is required. In the DeepConcrete transfer project, algorithms are being developed to segment and map the precast concrete warehouse on the basis of live camera images. The identification (recognition) of reinforced concrete components by automated image processing algorithms is particularly difficult due to very similar looking components (in colour and shape) and different weather and light influences. Traditional image processing algorithms are not able to cope with these challenges, which is why deep learning based methods are used in the project.

The activities in the transfer pilot will make a significant contribution to the automation and optimisation of the medium-sized construction industry. The work on AI-controlled component cataloguing and warehouse automation offers BREMER AG short-term added value and is at the same time part of the company's long-term innovation strategy.