Embedded systems can be found everywhere. But modern embedded systems get increasingly complex and are often operated under complex, uncertain or changing conditions. Our mission is hence to make them 'smart' and robust so that they autonomously adapt to the specific needs by themselves, but are still safe and engineered easily. We therefore apply different soft computing and organic computing techniques and combine them with online machine learning to simplify the engineering process and to increase the flexibility and robustness of the embedded system at runtime. Our research lines devide in four main areas: Handling of Uncertainties, Online Regression, Intelligent Control and Classification.
This research is complemented by works on architectures and practical applications. The later is very important for us because the usage of research results in practice leads to severe realistic constraints and more challenging requirements then pure simulations. We thus carry out cooperations with industrial partners (see Projects & Applications) and built and use several demonstrators (see Demonstrators & Robots).
As a consequence, a special emphasis is put on practical usability issues like safety and engineerability in order to mature the tools and methods and to make them easy to use in practice.
 |
 |
 |
 |