Central Car Server (CeCaS) for Highly Automated Vehicles Software Architecture Developed for Future Cars

Related Vendors

Technische Universität München (TUM) Lehrstuhl für Fahrzeugtechnik (FTM)

dataTec AG

Taiwan Machine Tool & Accessory Builders' Association (TMBA)

Wuxi InfiMotion Technology Co., Ltd.

To make autonomous vehicles as safe, affordable, and competitive as possible, researchers at the Technical University of Munich (TUM), together with partners from the automotive industry, have developed a centralized architecture for the software-driven vehicle of the future. The software largely generates itself, and various scenarios with autonomous vehicles can be tested in advance on a test bench.

To ensure that the cars of the future can navigate roads safely and reliably regardless of environmental conditions, vast amounts of data must be processed. This data comes live from sensors in the vehicle during driving and from databases and/or simulations on test benches during vehicle development.

"For autonomous driving, the data captured by the vehicle itself is combined with data from fixed cameras, lidars, or radars on overhead sign structures, or from other vehicles in the vicinity. That would be the maximum amount of information one could obtain," the Director of the TUM Chair of Robotics, Artificial Intelligence, and Real-Time Systems, Professor Alois Knoll, says."

Evaluate Data on the Spot

The purely software-based and centralized vehicle architecture, which evaluates and utilizes this data ad hoc, has been developed over the past three years by researchers at TUM and various partners from the automotive and chip industries as part of the "Central Car Server" (CeCaS) research project funded by the Federal Ministry for Research, Technology and Space (BMFTR). Such an architecture will be required for vehicle generations from 2033 onwards.

The Advantages of the New Vehicle Architecture Are As Follows:

• Scenarios can be tested realistically in simulations: Vehicles in reality are exposed to various traffic and weather conditions, which they cannot yet handle fully autonomously. To address this, the researchers have created a simulation environment in which diverse scenarios can be generated using powerful graphics chips. After training, the vehicle effectively has the knowledge for the respective situation "on board." Additionally, the scenarios can be made available to users from the automotive industry and research through an "open-source access."
• Centralized and standardized data processing drastically reduces costs: Traditional vehicles often use more than a hundred individual control units. Highly programmable high-performance computers, as in the CeCaS concept, will largely replace them in the future. This means that the need for extensive wiring between the control units is eliminated, assembly becomes simpler, and costs are reduced. More importantly, new functions will be possible purely through software upgrades. Additionally, software development can be customized by customers individually, similar to mobile phones.
• All functions can be tested on the test bench using a digital twin: At TUM's test bench, vehicles can be securely clamped with all their axles and wheels for testing. This allows not only the testing of driver assistance systems, anti-lock braking systems, or new emergency braking assistants. "Using a digital twin of the vehicle, we can also simulate scenarios that can then be tested 'live' on the test bench," explains TUM researcher Knoll. Additionally, scenarios can be simulated and trained in which past accidents involving autonomous or semi-autonomous vehicles occurred—without anyone being harmed.

Artificial Intelligence: Software is Created Effortlessly

For TUM Professor Knoll, a key advantage of the future vehicle architecture lies in its ability to accelerate development processes and thus innovation. As TUM research findings within the CeCaS framework show, software can be developed increasingly quickly using artificial intelligence and generative language models. Specifications are almost always available in text form, and these reflect the behavior of a technical device.

The TUM researchers have demonstrated that language models can process specifications, provided they are consistent, complete, and free of contradictions—something that AI can verify. This allows new software code to be created in seconds, essentially by design. However, this requires that the entire architecture of the vehicle is compatible. "Understanding cars as software-defined vehicles, or software platforms, is simply essential to remain competitive in the vehicle market in the future," summarizes Prof. Knoll. (sg)

Future-Proof Authentication with Universal RFID Readers

Universal RFID Readers: One Solution for All Credentials