A consortium of around 30 partners has completed a three-year effort to develop a centralised software architecture designed to support highly automated vehicles, with Germany’s Technical University of Munich (TUM) outlining key technical advances from the project. The Central Car Server (CeCaS) research initiative, launched in 2022 and led by semiconductor maker Infineon, aimed to create a high-performance computing platform that acts as the central processing unit for future software-defined vehicles.
The project was backed by a total budget of 88.2 million euros, including 46.2 million euros in funding from the German Federal Ministry of Research, Technology, and Space (BMFTR). Major industry partners included Volkswagen through its software unit Cariad, along with suppliers Bosch, Continental and ZF, while several Fraunhofer institutes and universities contributed on the research side. The central goal of the project was to develop an architecture capable of processing vast amounts of real-time driving data required for highly automated and autonomous driving functions.
See also: Nvidia Launches New Autonomous Driving AI Model to Advance ‘Physical AI’ Vision
“For autonomous driving, the data recorded by the vehicle itself is combined with data from permanently installed cameras, lidars or radar sensors on sign bridges or from other nearby vehicles. That would be the maximum amount of information you could get,” said Prof. Alois Knoll, head of the TUM Chair of Robotics, Artificial Intelligence, and Real-Time Systems. TUM researchers, working with other project partners, developed a fully software-based, centralised vehicle architecture designed to handle such data streams in real time. The architecture is expected to become relevant for vehicle generations from around 2033.
The new system enables the simulation of complex driving scenarios, including adverse weather conditions that continue to challenge current autonomous systems, using high-performance graphics processors. Once trained in these simulated environments, vehicles can retain the learned responses and handle similar situations autonomously in real-world operation. The CeCaS approach also replaces the hundreds of individual electronic control units used in conventional vehicles with programmable high-performance computers that can be updated through software.
See also: Raytron Rolls Out 180° Thermal Cameras for Autonomous Trucks
As part of the project, TUM integrated a Volkswagen ID.BUZZ provided by Cariad as a functional test vehicle. This allowed researchers to test real-world traffic scenarios using a digital twin in a live test-bench environment. “Using a digital twin of the vehicle, we can also import scenarios and perform live testing on the test bench,” Knoll said, adding that accident scenarios involving automated driving systems can be reproduced and corrected before deployment on public roads.
