Institute of Aircraft Production Technology

Description:

Hardly any product manufactured in Germany is as complex as the airplane. Several million components from various suppliers along the entire supply chain are required for a single aircraft. The highest demands are placed on the identification and documentation of components throughout their entire life cycle in order to guarantee complete traceability and quality control. The intelligent digitalization of these tasks requires state-of-the-art identification technologies that are used by all users in the value chain. The aim of the ILIdenT project is to develop and test such intelligent and aviation-compatible identification technologies. The challenge and opportunity of the project is the comprehensive enablement of digitalized identification processes for various end users.

Over a period of two years, the project developed AI-based solutions for identifying components. Since distinguishing similar components can require the processing of features in the micrometer range, sensor solutions were used to detect such features. The project partners not only rely on fully automated processes, but also consider the role of people in the identification process. In particular, trust in AI-based data processing has been addressed and created through Explainable Artificial Intelligence (X-AI) approaches.

Eleven partners from the Hamburg area came together to work on the project. From the suppliers Scholz Mechanik and Innovint to the industry giant Airbus and the maintenance company Lufthansa Technik, the company brings together representatives from the entire supply chain. The highly developed high-tech companies Synergeticon and 3d.Aero deliver state-of-the-art AI sensor technologies. The Artificial Intelligence Center Hamburg supports the certification of AI solutions. The Institute for Product and Production Management at the Hamburg University of Applied Sciences and the Institute of Aircraft Production Technology at the TU Hamburg are part of the consortium as scientific support. As the largest independent association of suppliers and service providers in the aviation industry, Hanse-Aerospace brings the project results to the supplier industry. The service companies Orange Engineering and Capgemini complete the venture with their expertise in the aviation sector.

The project is supported by the Hamburg Ministry for Economics and Innovation and the Hamburgische Investitions- und Förderbank.

Synthetic training data for AI applications

Visual object identification should increasingly be used for the comprehensive digitalization of production processes. In line with the current trend, neural networks are used as the basic architecture for AI object identification. Such neural networks must be trained with targeted data in order to map the diverse and specific process variants in the industry. These large amounts of data are usually not available in companies and manual generation of labeled data is uneconomical. The IFPT is therefore developing methods for generating AI training data for industrial processes. As part of the DEPOT and ILIdenT projects, a toolbox for generating synthetic training data is being developed developed with which individual data sets can be generated for visual applications in intralogistics.

Contact person at the institute: M.Sc. Daniel Schoepflin