MANNHEIM-CeCaS

Central Car Server – Supercomputing for Automotive

Project Information

Funding:
Federal Ministry of Education and Research (BMBF)
Electronics and Software Development Methods for the Digitalization of Automobility (MANNHEIM)

Runtime:
December 2022 – November 2025

Consortium:
28 Partner, u.a. Infineon, BOSCH und KIT

Project Goal

The MANNHEIM-CeCaS project aims to develop an automotive supercomputing platform to address the massive data processing and complex computational needs of highly automated vehicles. At its core, the project focuses on creating a central car server utilizing cutting-edge automotive-qualified high-performance processors based on FinFET technology. This platform will be enhanced with application-specific hardware accelerators and an adaptive software system tailored for autonomous vehicles.The research scope encompasses innovative approaches such as folded neural networks and event-based neuromorphic accelerators. Additionally, the project will explore necessary adaptations to on-board networks and automotive-grade assembly and interconnection technologies. The consortium's ultimate goal is to achieve full automotive qualification (ASIL-D) at the system level, ensuring the highest safety standards for autonomous driving applications.

Contribution from the Chair of AIR

Within the CeCaS project, we as TUM Chair of AIR are focusing on two key areas aimed at revolutionizing automotive development workflows. The primary goal is to implement an automated toolchain that translates textual requirements into executable code while ensuring compliance with automotive standards. By leveraging cutting-edge Large Language Models (LLMs) in synergy with traditional model-driven engineering, our solution accelerates vehicle software development and reduces the time and effort needed for standardization and code generation. In addition, we are developing a flexible testbench that integrates physical vehicles with a simulation environment, enabling rapid prototyping for autonomous driving systems. This reduces both the cost and complexity associated with testing and development, ensuring quicker adoption of research outcomes in the automotive industry.

Researchers


no image

Dott. Mag. Nenad Petrovic


no image

M.Sc. Nils Purschke


Picture of Sven Kirchner

M.Sc. Sven Kirchner


Foto von Fengjunjie Pan

M.Sc. Fengjunjie Pan


kein Bild

Krzysztof Lebioda


Picture of Vahid Zolfaghari

Vahid Zolfaghari


Foto von Andre Schamschurko

Dipl.-Inf. Andre Schamschurko