The Development Community Needs to Step Up

At the prostep ivip Symposium in Berlin, it became evident that the German automotive industry is under considerable pressure. “The situation is really, really challenging,” admitted Thomas Kamla, CTO at Volkswagen, responsible, among other things, for the development of the ID.1: “We need to increase our speed and drastically reduce our costs. That means we must work intensively on our methodology, our processes, and our toolchain – because that is the foundation of our future efficiency,” said Kamla, who is also a member of the prostep ivip board.

Much hope is being placed on the SDV and the digital, AI-based twin that is expected to significantly reduce development effort. In the light of these dramatic changes, the prostep ivip Association also announced a realignment of its strategy. “The situation has changed dramatically, not only geopolitically, but also with regard to software-defined products and advances in artificial intelligence,” noted Tomohiko Adachi of Mazda. The association aims to serve as a beacon throughout this transformation.

VW says goodbye to the Global Car 

Thomas Kamla reported that Volkswagen plans to move away from the concept of a Wolfsburg-developed 'Global Car' with identical design and technology for nearly every market. “We will shift competencies and processes to our development centres around the world,” explained Kamla. The target is to save €200 million in R&D across the group. To move away from the costly maintenance of over 450 systems, Volkswagen has opted for a platform approach. The company plans to use 3DX as its PLM system and data backbone. “The plan is to use it mainly out of the box, with minimal customisation,” Kamla said – a remark that drew laughter, as PLM is widely regarded as a particularly difficult area to standardise.

Volkswagen wants to establish a “new normal” in which the focus lies on the CDV – customer-defined vehicle – in combination with the SDV. For the ID.1, the development process was shortened by 30 percent. On the one hand, this was achieved through an increased use of virtualisation technologies, which at VW are referred to as Road2Rig2Byte. Especially in testing, the aim is to transition towards virtual testbeds and simulation. On the other hand, each development project now begins with a specific focus area, based on concrete customer personas and their individual needs, for which data is collected beforehand.

Will AI put Germany back in the lead? 

Kamla presented several personas: a 19-year-old who frequently uses social media, a care worker who uses the car professionally, and a 62-year-old DIY enthusiast. This suggests that Volkswagen – unlike its Asian competitors – intends to further pursue the highly individualised path even for vehicles priced around €20,000. Many questions remain, however: what happens when customer needs or professions change? How many 19-year-olds use the car’s entertainment system rather than their smartphones?

At least one consensus was reached at the event: AI is the antidote to increasing complexity. According to Sabine Scheunert, Managing Director Eurocentral at Dassault Systèmes, AI-based virtual twins “could bring the German automotive industry back to pole position.” The sector is undergoing a “massive revolution.” The technology relies on large language models and provides developers with a “virtual companion” that can take over a significant share of time-consuming tasks. Dassault believes that innovation cycles can be reduced from around seven years to less than three. Scheunert also highlighted the importance of digital sovereignty, emphasising that it is vital for European companies to keep their data within Europe.

Data must be structured 

The greatest challenge – unsurprisingly – is data quality, which is essential for AI to be effective. This becomes even more critical as agent-based AI takes on increasing responsibilities within the development process. “Ontology” – a formal description of knowledge in a specific domain – was therefore one of the most frequently heard terms at the Berlin conference. “We are all sitting on a treasure chest of data – but we also all know: that treasure chest is often very hard to unlock,” said Ulrich Wolters of Bosch Connected Industry. Information must be brought together from a semantic perspective so that AI models can actually work with enterprise data.

According to the experiences of Bosch, Mercedes, and Schaeffler, the ontology provided by Catena-X and the administration shell used for it offer a solid foundation for the necessary interoperability. The decision to align the digital twin with the administration shell now enables Bosch to take part in collaborative scenarios within Catena-X, said Wolters.

Paying for the Future

The major automotive players are investing significant sums in re- and upskilling their people. With increasing use of AI, career paths are changing, and the right further training is becoming even more crucial. This is especially challenging for medium-sized companies.

Daniela Hoffmann

Experience shows: if AI is applied directly to raw data, the result is only about 20 percent accurate. When semantic, structured data is used, accuracy rises to 60 percent. “We are trying not to train the AI. Agent-based systems provide environmental context to avoid hallucinations in large language models. We guide these systems to navigate the data. This is a significantly more cost-efficient approach than training,” explained Wolters.

According to Sebastian Handschuh, Head of Catena-X Architecture and Operation at Mercedes-Benz, far too much work is still being done manually in data preparation: “When we prepare a dataset manually and send it off now, it takes several days to weeks before these feedback cycles are closed.” This applies to areas such as demand and capacity management, as well as CAD and geometry data. Since millions of files are received monthly through the sales and partner network, the digitalisation potential is immense. To unlock this potential, Handschuh believes a shift in mindset is needed: companies should no longer view themselves only at their individual corporate level, but – as demonstrated at the Berlin event – embrace the idea of a trustworthy community and move beyond the traditional OEM-supplier mindset.

This article was first published at automotiveit.eu