Materials Center Leoben Forschungs GmbH
Materials Center Leoben Forschung GmbH (MCL) is an international research company working in the field of materials research and technology. MCL specialises in materials, their manufacture and processing, as well as innovative material applications.
At MCL, more than 140 highly qualified employees work together with over 140 industrial and scientific partners on fundamental and innovative developments along the entire value chain, from materials synthesis and processing through to in-service behaviour.As the operating company of the COMET K2 Competence Center “MPPE – Integrated Research in Materials, Processing and Product Engineering”, MCL is the ideal partner when it comes to demanding and complex, interdisciplinary research and development tasks.
MCL’s core service offerings include characterising materials and components with respect to their structure and microstructure, determining the mechanical and physical properties of materials, developing material models for simulations as well as advanced simulation models , damage analysis, and advice on the choice of materials. MCL’s key advantage lies in its combination of experimental laboratory analysis with calculations and simulations, state-of-the-art technical facilities, and wide-ranging specialist knowledge of the most diverse range of materials. MCL has the expertise and experience required to provide scientifically sound results and targeted support in practical material and product development. Current research topics on experimental characterisation and numerical simulations of materials range from the nano-scale up to the scale of large products such as railway tracks and gas pipelines.
MCL has the expertise and experience required to provide scientifically sound results and targeted support in practical material and product development.
MCL’s role is that of a broker and activator between machine learning and real-time implementation on specialized hardware (Smart Cameras, PLCs). While training of models typically focuses on classification performance, actual implementation on embedded hardware is time consuming and often bound to hard real-time requirements.Our expert system will enable the two groups to work together more effectively and to observe each other’s requirements. Specifically, execution time estimation for selected target platforms will allow for a second ranking criterion in addition to classification performance. Automatic code generation for specific target platforms will sharply reduce the time from model training to working system for selected types of classifiers. This in turn should lead to more iterations and more intelligent and robust classifiers used in semiconductor production.
- an expert system able to (i) estimate execution time and (i) generate code for complex classifiers on different computing platforms relevant to semiconductor production facilities (WP4).
- feedback to WP3 (machine learning) concerning the execution time of developed classifiers and execution time-relevant classifier parameters.
- Implementations of classifiers selected for use in demonstrators (WP5).