High-tech Embedded Systems Challenges (CONDOR)
Electron microscopes are highly complex devices whose design combines quite a number of disciplines: (high-voltage) electron physics, electromagnetism and electron optics, image sensors and image processing, mechatronics, instrument conditioning (vacuum, heat), electronics (amplifiers, sensors, embedded control) and software engineering. As such, they require a significant amount of configuration by expert-operators. The need arises, however, to integrate electron microscopes in production processes. The challenge is therefore to combine high performance with ease of use, flexibility and self-automation/configuration.
DistriNet's involvement in the project focuses on the software that controls the electron microscope. In order to develop software for such highly complex embedded systems, key concerns that need to be dealt with are predictability, performance and the traceability of features through the entire development cycle. Instead of building and rebuilding prototypes, state of the art software modelling techniques will be used. This includes research for well suited dedicated modeling languages as well as the integration of existing formal methods (which have previously only been used on academic projects) on an industrial-size case studies which will deal with a variety of problems such as:
- Auto-focus: A system automatic focus and correction for astigmatism that is effective for a wide variety of samples will be developed for scanning microscope systems.
- Positioning: Using combined beam and sample actuation, it is possible to obtain quicker, more accurate, more stable and robust positioning of the sample.
- Self-optimizing imaging chain: Using criteria such as throughput, precision, and fit for metrology, the controllable parameters of the image formation chain will be used to create an optimal image.
- Quantitative measurements and auto-calibration: The quantitative measurement performance of the electron microscope will be enhanced using models for design improvements and will be made part of the instrument control. As such, the accuracy, repeatability (between measurements and between instruments) and stability (over time) of sample data can be predicted and improved.
Additional topics of research that will possibly be tackled in this project are model transformations and semi-automated architecture reconstruction.