Infineon and partners present supercomputing platform

Vehicles are becoming increasingly connected, intelligent, and autonomous, requiring powerful computing architectures with multi-gigabit interfaces. 28 research partners from industry and academia have now jointly presented a new supercomputer 'Made in Germany', the demonstrator of a central computing platform for the automotive industry, which meets the highest safety and reliability standards. Under the leadership of Infineon Technologies, the partners have developed the platform over the past three years as part of the Mannheim-CeCaS (Central Car Server) research project. The project was funded by the Federal Ministry for Research, Technology and Space (BMFTR) and, with a project volume of 88.2 million euros, is one of the ministry's largest national automotive research projects. The demonstrator, integrated into a modern electric car, was presented at the project's closing event in Munich and proved its functionality there.

Pony.ai launches affordable Gen-7 robotaxis

Pony.ai has commenced operations of its Gen-7 robotaxis in China. According to the company, the autonomous driving kit of this generation is 70 percent cheaper than before - a step intended to advance the scaling of driverless mobility.

Martin Large

"With the results of the Mannheim-CeCaS project, we are taking a significant step towards the mobility of the future - this is where research and innovation hit the road," says Frank Badstübner, spokesperson for the Mannheim-CeCaS project management, and adds: "The energy and cost-efficient high-performance computing platform developed jointly in the consortium is standardised, modular, and scalable. This impressively shows how we, with excellent partners from business and academia, significantly contribute to the innovative strength of the German automotive industry in global competition."

Centrally controlled and modular for many vehicle types

The aim of the Mannheim-CeCaS project was to develop a powerful and real-time capable supercomputer for autonomous driving from level 3 to 5. The project consortium worked on the design of processors, interfaces, and system architectures. A flexible software environment was specifically tailored to the requirements of the latest algorithms in the automotive sector, particularly for the use of artificial intelligence (AI). The CeCaS Mannheim team consolidates the vehicle's intelligence in a central hardware-software architecture, instead of using many separate control units. This leads to less complexity and more efficiency, while the platform itself remains flexible and can easily adapt to new requirements. Future modules could be easily retrofitted without having to redevelop the entire vehicle. This central solution enables faster developments and easier adaptations for different vehicle sizes, performance classes, and functions. 

Bertrandt presents autonomous shuttle for rural areas

Together with the Bern University of Applied Sciences, the engineering service provider has developed the BumbleB shuttle to primarily provide answers to current mobility issues beyond urban areas.

Götz Fuchslocher

The different modules are simultaneously regulated via a computing node. This includes safety-critical systems such as the engine, transmission, and brakes, as well as on-board cameras, parking aids, temperature and distance sensors, motors for window lifters and seat adjustment, or air conditioning and on-board entertainment. Thanks to the new zonal architecture, the vehicle also becomes lighter, as fewer cables are needed. And it reduces energy consumption, which in turn positively affects the range of electric vehicles. The communication between the components takes place in real-time. The research team relied on Ethernet-based network technology for this. Despite the huge data volume, the Mannheim-CeCaS system remains extremely flexible and scalable. Additionally, the vehicle can be updated via WLAN, eliminating the need to visit the workshop for updates. In a cost-optimised rapid prototyping approach, the team managed to implement and validate the zone-based demonstrator in just nine months. The complete automotive qualification at the system level was prepared. The insights gained can also be incorporated into modular chiplet technologies or RISC-V-based application processors.

This article was first published at automotiveit.eu

Sponsored

automotiveIT car.summit

SV Tower 11-12 November 2025, Munich