Co-funded by Bundesministerium für Wirtschaft und Energie (BMWi)
Duration: April 2016 – September 2018
The M3D project aims at improving currently established workflows handling spare parts for specialized industrial machines by exploiting recent advancements in the area of 3D reconstruction, object identification and 3D printing.
Any spare part service attempts to achieve: “Getting the right part to the right place at the right time and the right cost”. In the case of specialized and complex mechanical appliance (e.g. train or car) this process typically involves highly qualified expertise for the unique part identification, for potentially required engineering steps and for logistics. Nevertheless, customers demand immediate response and solution to their needs when maintaining or repairing such devices. Therefore, the objective is to accelerate the spare part identification, ordering and delivery by improving the major parts of the workflow:
- Fast, easy-to-use and accurate on-the-spot 3D scan and reconstruction of the required spare part using a mobile solution.
- Automated, easy-to-use and reliable on-the-spot part identification using the previously acquired 3D data by searching through a CAD-database.
- Improved methods for manufacturing (e.g. 3D printing) the required spare parts by using available CAD data or the 3D scan data.
Fraunhofer HHI is responsible for the project management and is furthermore involved with two groups. The Computer Vision and Graphics group will contribute their expertise on multi-view 3D reconstruction and will create an image-based mobile 3D reconstruction workflow for the subsequent identification of the object in large databases. The reconstruction process will especially address the complex capturing environment as well as challenging object characteristics. While the acquisition will be performed on the mobile platform, computationally demanding tasks will be moved to the cloud. In order to assist the user during the acquisition, a mobile user interface will be developed that provides the user with visual feedback on the one hand and allows him to provide additional information guiding the reconstruction process on the other. The Embedded System Group will develop a computing platform based on heterogeneous processing architecture concepts incorporating HW accelerators to process the 3D reconstruction in the cloud.