Sign in with
Forget Password?
Learn more
share this!
Share
Tweet
Share
Email
December 1, 2025
by Andreas Schmitz, Technical University Munich
edited by Sadie Harley, reviewed by Robert Egan
scientific editor
associate editor
This article has been reviewed according to Science X’s editorial process and policies. Editors have highlighted the following attributes while ensuring the content’s credibility:
fact-checked
peer-reviewed publication
trusted source
proofread
To make autonomous vehicles as safe, affordable and competitive as possible, researchers at the Technical University of Munich (TUM) have teamed up with partners from the automotive industry to develop a centralized architecture for the software-controlled vehicle of the future. The software is largely self-generating and allows the advanced simulation of any scenario involving autonomous vehicles on a test bench.
To ensure that the cars of the future can travel safely and reliably on roads regardless of environmental conditions, vast amounts of data must be processed. The data are collected in real time from sensors in the vehicle while driving and from databases and/or simulations on test benches during vehicle development.
“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,” says Knoll, head of the TUM Chair of Robotics, Artificial Intelligence and Real-Time Systems.
Over the past three years, researchers at TUM and various partners from the automotive and chip industries have developed a suitable vehicle architecture that evaluates and uses the data on an ad-hoc basis—as part of the “Central Car Server” (CeCaS) research project.
The research is published in the 2025 IEEE Intelligent Vehicles Symposium (IV), the journal IEEE Access and the preprint server arXiv
A centralized and entirely software-based architecture of this kind will be required for vehicle generation from 2033 onward.
The advantages in detail:
In reality, vehicles are not yet fully capable of handling many traffic and weather conditions that they encounter. To address this, researchers have created a simulation environment in which a wide range of scenarios can be generated with the aid of powerful graphics chips. After training, the vehicle has the knowledge to cope with the given situation “on board.” The scenarios can also be made available to users from the automotive industry and research via open-source access.
Conventional vehicles often use more than a hundred individual control units. Versatile, programmable high-performance computers such as those used in the CeCaS concept will largely replace them in the future. This will eliminate the need for many connecting cables between control units, make installation easier and reduce costs.
Above all, however, it will be possible to add new functionality purely through software upgrades. And, as with mobile phones, the software development can be customized by customers.
The TUM test bench allows vehicles to be securely clamped in place with all axles and wheels for testing. This means that not only driver assistance systems, anti-lock braking systems and new emergency braking assistants can be tested.
“Using a digital twin of the vehicle, we can also import scenarios and perform live testing on the test bench,” explains TUM researcher Alois Knoll.
In addition, scenarios from real-world accidents involving autonomous or semi-autonomous vehicles can be imported and used for training—without anyone being harmed in the process.
For TUM Professor Knoll, a key advantage of the future vehicle architecture is that it will accelerate development processes and thus innovation as well. As TUM research results within the framework of the CeCaS project show, software can be developed ever more quickly with the help of artificial intelligence and generative language models. Specifications are almost always available in text form. And these reflect the behavior of technical devices.
TUM researchers have shown that language models can process specifications as long as they are consistent, complete and free of contradictions, which in turn can be checked by AI. This allows new software code to be created in seconds, virtually by design. However, this requires the entire architecture in the vehicle to be compatible.
Knoll said, “Understanding cars as software-defined vehicles, i.e. software platforms, is simply necessary in order to remain competitive in the vehicle market in the future.”
More information: Sven Kirchner et al, Generating Automotive Code: Large Language Models for Software Development and Verification in Safety-Critical Systems, 2025 IEEE Intelligent Vehicles Symposium (IV) (2025). DOI: 10.1109/iv64158.2025.11097503
Krzysztof Lebioda et al, Are Requirements Really All You Need? Using LLMs to Generate Configuration Code: A Case Study in Automotive Simulations, IEEE Access (2025). DOI: 10.1109/access.2025.3597748
Nenad Petrovic et al, GenAI for Automotive Software Development: From Requirements to Wheels, arXiv (2025). DOI: 10.48550/arxiv.2507.18223
Explore further
Facebook
Twitter
Email
Feedback to editors
5 hours ago
0
Nov 28, 2025
0
Nov 27, 2025
1
Nov 27, 2025
0
Nov 25, 2025
0
1 hour ago
3 hours ago
3 hours ago
5 hours ago
9 hours ago
Nov 30, 2025
Nov 29, 2025
Nov 28, 2025
Nov 28, 2025
Nov 28, 2025
Apr 25, 2023
Sep 13, 2022
Mar 3, 2024
Jun 17, 2025
Aug 14, 2025
Jun 19, 2024
1 hour ago
Nov 21, 2025
Nov 20, 2025
Nov 19, 2025
Nov 18, 2025
Nov 18, 2025
A centralized, software-based vehicle architecture enables real-time, ad-hoc data analysis and advanced simulation of autonomous driving scenarios. This approach reduces hardware complexity and costs by replacing numerous control units with high-performance computers, supports rapid software updates, and allows comprehensive testing via digital twins. AI-driven code generation further accelerates development.
This summary was automatically generated using LLM. Full disclaimer
Use this form if you have come across a typo, inaccuracy or would like to send an edit request for the content on this page. For general inquiries, please use our contact form. For general feedback, use the public comments section below (please adhere to guidelines).
Please select the most appropriate category to facilitate processing of your request
Thank you for taking time to provide your feedback to the editors.
Your feedback is important to us. However, we do not guarantee individual replies due to the high volume of messages.
Your email address is used only to let the recipient know who sent the email. Neither your address nor the recipient’s address will be used for any other purpose. The information you enter will appear in your e-mail message and is not retained by Tech Xplore in any form.
Daily science news on research developments and the latest scientific innovations
Medical research advances and health news
The most comprehensive sci-tech news coverage on the web
This site uses cookies to assist with navigation, analyse your use of our services, collect data for ads personalisation and provide content from third parties. By using our site, you acknowledge that you have read and understand our Privacy Policy and Terms of Use.
