PY2DLL

PY2DLL Abbildung

Researcher: Lohannes Leuschner
Project funding: Europäischer Fond für regionale Entwicklung (EFRE)
Project sponsor: Bremer Aufbaubank GmbH (BAB), funding no FUE0656B
Partners: Sebastian Schmale, SIKORA AG
Duration: 06.02.2023 - 31.10.2023

Logos

The purity of products and the related quality assurance play a major role in the development of production. To maximize savings of material and costs, it is necessary to reliably measure the finished products as well as to detect abnormalities with respect to the given product specification. The analysis of inhomogeneous, extruded products, such as corrugated pipes, with increasing geometric complexity requires the development of new evaluation algorithms, with an increasing usage of artificial intelligence (AI). AI-based methods often are subject of ongoing research and are typically developed in a rather academic environment with high-level programming languages such as Python. The project PY2DLL, which is funded by the European Union, targets a largely automated strategy for porting a research code-base written in Python into a dynamic-link library (DLL) that can be used by an existing control application written in C++. Limited resources and efficiency requirements need to be taken into account in doing so. This ensures the applicability of the newly developed AI-based analysis software in practice.

Besides fulfilling the efficiency requirements, which are imposed by the hardware specifications of the measuring devices and the targeted evaluation frame rate, maintainability and testing are also elementary goals of this project. More precisely, to achieve maintainability, the Python code needs to be ported or integrated into the DLL in an almost completely automated manner, such that continued developments in Python can be transferred into practice with a reasonable amount of work. This particularly excludes the possibility of re-implementing larger portions of code in C++. By thorough and also automated testing, it is assured that the functionality, which has already been tested during development, is retained without modification in the transfer from Python to C++.

The industry partner SIKORA is one of the leading companies for measurement and control technology as well as inspection, analysis and sorting systems, and one of the hidden champions among German mid-sized companies. In the cable industry, SIKORA is world market leader for measurement technology. The project PY2DLL seeks to implement a software innovation based on artificial intelligence (AI) in practice, which will significantly extend the applicability of the measuring devices of the SIKORA AG.