The German Federal Ministry for Economic Affairs and Energy (BMWi) is funding German AI research with 11.2 million euros to develop safe, innovative electronic systems for future generations of vehicles. In the project progressivKI, eighteen top-class partners are working together in a consortium under the coordination of Robert Bosch Multimedia Car GmbH in Hildesheim and with project management support by edacentrum. One of these partners is Fraunhofer ENAS with the department Advanced System Engineering (ASE) in Paderborn.
The aim of the project, which started on April 1, 2021, is to develop a generalized AI-supported design process for electronic systems in vehicles through the cooperative operation of AI systems to develop electronic systems faster and more reliably and thus to claim a globally effective innovation and technology leadership of the participating companies for future key technologies. The planned work also considers the current GAIA-X developments of the BMWi.
Currently, no AI-supported design is carried out in the development process of vehicle components. Manual procedures and semi-automatic processes are often used in layout and circuit diagram design and in the development of wire harnesses. In addition, from the point of view of developing functionally safe electronic systems, the properties and interaction of the individual components and the subsystems require a holistic view of the system properties. In the future, this can only be ensured with the help of AI-supported methods and tools in the design process, given the increasing complexity of the systems and the increasing specialization of the development groups.
The project progressivKI intends to change this by developing a modular AI-based platform to support and accelerate upcoming electronic systems in the automotive sector. The planned AI modules are intended to support developers in such a way that computing times are optimized and simulation results can be interpreted faster and in a more targeted form as well as the results can be reused in other variants. The platform is being designed for use throughout the entire development process so that all interactions during the process can be considered.
The implementation of the AI modules requires extensive training data in numerous areas of electrical engineering. Fraunhofer ENAS supports the consortium in generating and applying these data. For this purpose, simulation results for various use cases in the automotive sector are to be generated automatically by electromagnetic simulation platforms. These data will then be pre-processed and classified to form the specific AI modules. In addition, Fraunhofer ENAS will also implement their expertise into the modules to reduce the required amount of training data and at the same time to maintain the correct operation of the modules. To ensure the functionality in practice, a verification of the AI modules and an adaptation of the training data to the real operating conditions will also take place.
Under the leadership of Robert Bosch Car Multimedia GmbH, the partners Infineon Technologies, Zuken, Microchip Technology Germany, Celus, Luminovo, Hood, Cloud&Heat Technologies, Binder Elektronik, DIQA Projektmanagement, EMC Test NRW, InnoZent OWL e.V. as well as Fraunhofer FIT and Fraunhofer ENAS, FZI Forschungszentrum Informatik, TU Dortmund, HSU Hamburg, TU Berlin and HS Hamm-Lippstadt/System Integration Laboratory are participating in the three-year project, covering all necessary components of the value chain.
Funding code 19A21006A-R