Construction companies that manufacture precast reinforced concrete components for industrial and commercial construction have extensive outdoor warehouses where these components are temporarily stored and held. Current warehouse mapping systems and warehouse management processes are largely based on manual processes, which leads to inefficient placements of the precast parts and thus to a warehouse that is not optimally utilized. The transfer project "DeepConcrete" aims to automatically map BREMER AG's outdoor warehouse using camera systems and AI methods. For this purpose, image recordings from cameras on the gantry crane are to be segmented. The automated recording of the warehouse inventory thus enables the implementation of optical monitoring and AI-based warehouse management. The long-term vision of the project is a digital platform for order and goods flow management as well as a digital mapping of the warehouse to optimize customer offers, logistics and production planning.

Storage of precast reinforced concrete parts is organized manually at the producer's premises

BREMER AG (headquartered in Paderborn, Germany) produces all precast reinforced concrete elements required for building construction in its own plants with connected large-scale external warehouses. This enables a guarantee of competitively short construction times and sustainable use of production capacities. Currently, storage is organized manually by employees. Access to the precast elements is often time-consuming, as there is no optimal allocation of storage space. As a result, resources such as storage space, working time and energy are not used sustainably and there is scope for optimizing warehousing. Conventional automated warehouse management is difficult to implement due to the material properties of reinforced concrete and the variety of precast elements. Here, cataloging and segmentation of the warehouse via image recording by camera systems attached to gantry cranes offers a promising approach to building an intelligent warehouse management system.  

AI algorithms for automated warehouse mapping and management.

The vision of DeepConcrete is to create a digital platform in the construction industry for order and flow management. The main objective within the project duration is to develop image-processing AI solutions for automated recognition of precast reinforced concrete elements and warehouse mapping. Subgoals include image data acquisition of the warehouse and generation of training datasets. In addition, potential AI algorithms will be identified and analyzed for their suitability for transfer learning. The selected AI algorithms will be trained and evaluated.



Further goals include the development of AI algorithms for automated warehouse management. This includes visual monitoring and documentation of warehouse inventory. Based on the requirements and tasks defined in the project, AI algorithms for warehouse management will be researched, summarized in a solution catalog, and implemented and evaluated for AI-based warehouse management.  

AI-based optimization of warehouse management and material flow planning.

The result of the transfer project is a solution catalog of AI-based algorithms for automated warehouse mapping and management. Concrete partial results include literature analyses, feasibility studies, solution concepts and their project-related evaluations. Companies can draw on the project results to automate their inventory management in an AI-based manner and to optimize their warehouse management with regard to logistical target variables. The resulting more efficient warehouse management leads to more cost-effective and sustainable processes for the regional construction industry.  



In the long term, DeepConcrete paves the way to a digital platform for order and goods flow management. With a digital mapping of the warehouse, customer quotations, logistics and production planning can be optimized. Extending automated warehouse management to additional external warehouses enables companies to plan material flows across locations and holistically.