Distributed and efficient data processing in digital radar networks for fully automated driving

The goal of the VERANO project is to research distributed, efficient data processing including AI methods in digital radar networks for fully automated driving. The focus is on the extensive digitization of all system-relevant functionalities as well as the development of an application-specific optimal computational load distribution between network controllers, sensor nodes (sensor edge) and central computer. By using state-of-the-art RFSoC FPGAs in the sensor nodes, different computational steps and AI algorithms are to be implemented during signal processing and their results are to be sent to high-performance central computers in a TSN (Time Sensitive Network). In the simplest case, pre-processing can consist of lossless compression of the raw data and can extend to complex AI-based procedures using a back channel to the central computer. The optimization of the computational load distribution between sensor nodes and central computer should be done with respect to application requirements regarding "image quality", cost, energy efficiency, reliability, real-time capability.... By optimizing the resource efficiency, a significant contribution to sustainability and the Green Deal will be made. The results are also transferable to other sensor types (camera, lidar, etc.) and applications such as Industry 4.0, logistics, medical technology.

Partners

Infineon Technologies AG, Robert Bosch GmbH, Mercedes-Benz-AG, Missing Link Electronics GmbH, KPIT Technologies GmbH, Fraunhofer Institut IPMS, Universität Ulm, Ruhr-Universität Bochum, TU Braunschweig, Universität Kassel