The development process for integrated sensor systems is very complex. Specialists such as analogue and digital circuit designers, verification and test engineers, software developers and others work closely together there. Design errors must be avoided as much as possible, as iterations and redesigns involve significant additional costs and time delays.
We are exploring the use of artificial intelligence to make the development process of integrated sensor systems safer and more cost-effective.
AI can support developers at many points in the process to avoid errors and apply informal knowledge that is difficult to map procedurally in an automated way. To achieve this, we are working on machine learning approaches whose algorithms learn, for example, the good design practices of experienced developers and provide valuable advice to even less experienced colleagues. In this way, review processes that are common practice in the design of integrated systems can be made more effective or even replaced.
AI also helps us to create fast simulation models. For example, we use neural networks to integrate non-functional properties into existing behavioural models very effectively. This in turn results in increased design reliability.
Artificial intelligence is also the key to evaluating large amounts of data, as is often the case when testing integrated circuits. Here we are working with AI as well to significantly relieve increasingly scarce personnel resources.
Eric Schäfer, M. Sc.
Head of Microelectronics / Branch Office Erfurt
eric.schaefer(at)imms.de+49 (0) 361 663 25 35
Eric Schäfer and his team research Integrated sensor systems, especially CMOS-based biosensors, ULP sensor systems and AI-based design and test automation. The results are being incorporated into research on the lead applications Sensor systems for in-vitro diagnostics and RFID sensor technology. It will assist you with services for the development of Integrated circuits and with IC design methods.
Against hacker attacks: Innovative chip architectures, modelling and verification methods for trustworthy electronics
We are developing copy protection solutions for integrated circuits to secure the know-how of partners
IMMS develops novel designer assistance system and smart models to design sensor and control chips for I4.0 applications
Christian Paintz, Melexis
“Particularly in the evaluation of measurement data, IMMS has impressively demonstrated that a learning algorithm is on a par with manual evaluation – while saving a great deal of time. We are also continuing to pursue methods for circuit and layout analysis, as we see great research and application potential here as well.”
Dr.-Ing. Dirk Nuernbergk, Melexis
“The institute has developed and implemented a program that automatically finds and evaluates critical parasitics even at the design stage. In consequence, layout optimisation, which normally takes so long, is massively accelerated. In a very short time, we were able to identify the problematic points in three circuits.”