What building blocks does a data and services ecosystem need for AI applications in the automotive sector? And how can the Gaia-X open IT infrastructure be used as the basis for this? The Gaia-X 4 KI project has spent the last three years researching the answers to these questions. The 13 participating research institutions and industrial companies have set up a prototype data space and trialled it in the areas of production and automated networked driving. Led by the German Aerospace Center (DLR) and funded by the Federal Ministry for Economic Affairs and Climate Action (BMWK), it is the first project in the Gaia-X 4 Future Mobility project family to present its results.
Project coordinator Dr. Sascha Knake-Langhorst from the DLR Institute of Transportation Systems summarises: "Only with an infrastructure that allows us to exchange data quickly and securely can we effectively apply artificial intelligence in the automotive sector. With our project, we have shown that Gaia-X, as a decentralised data infrastructure with its open, transparent and secure interfaces and standards, can provide the right basis for a wide range of applications."
At the start of the project, the researchers identified the technical and methodological building blocks they needed for a prototype data space solution. On a technical level, they decided to use the building blocks of the EDC (Eclipse Dataspace Components). On this basis, the consortium researched and developed various prototype data space solutions. These were tested in practice using nine different use cases from the fields of production and automated networked driving.
Data security plays a particularly important role in data transmission in automotive production. This applies both across company boundaries and within multinational corporations. The researchers therefore analysed the transmission of sensitive production data from manufacturing to partner companies that use it to train AI algorithms for quality control. As a result, they were able to determine that a Gaia-X-based ecosystem is well suited for this due to the shared basis of trust and thus opens up new optimisation opportunities in production. They also identified other strengths, such as the speed advantage of concluding digital contracts and the ease of finding subcontractors or customers in a shared data room.
For AI applications in automated and connected driving, the transmission of large amounts of data is an everyday challenge. For example, developers of automated driving functions use camera data to train AI algorithms, which must be anonymised to ensure data protection. In the Gaia-X 4 KI project, the scientists have tested how a fast and trustworthy exchange of large amounts of data between different companies can be carried out on the basis of Gaia-X. Another relevant use case in the ongoing operation of automated, connected vehicles is the streaming of data. For example, the functionality of automated vehicles such as shuttles can be monitored. The researchers have explained the corresponding results in more detail in ATZelektronik (5/2024). The topic of cyber security is also becoming increasingly important, as the constantly growing number of automated vehicles must be protected against attacks. The project was able to develop initial prototype solutions based on EDC for both requirements.
Gaia-X 4 KI already passed on its findings on the development of a Gaia-X-based infrastructure to the five sister projects in the Gaia-X 4 Future Mobility project family during the course of the project. For example, these projects orientated themselves on the early development of simple demonstrators (so-called minimal examples) and also relied in part on the Eclipse Dataspace Components. Core components of the dataspace solution and technical solutions were also adopted by the other projects. This promoted continuous exchange and accelerated the work in all projects. Gaia-X 4 KI's research into the legal and regulatory framework of a data and services ecosystem also serves as a cross-project basis. However, the ongoing projects in the project family will continue to work on their respective research priorities until 2025.
Videos of the individual applications are also available on the homepage: https://gaia-x4ki.eu/en/Demo
