Ulrich Eichhorn heads up Volkswagen Group’s R&D operations, and discusses impacts for vehicle development from artificial intelligence and electrification, and explanins how digital tools and data are changing engineering processes.


Eichhorn: We want to control the driving system (Photo: VW)

Volkswagen has to master a balancing act: On one hand, the carmaker needs investment discipline. On the other, there is tremendous pressure to innovate. Ulrich Eichhorn, who heads the VW Group research and development operations, talks about current efforts to electrify VW’s portfolio and breathe artificial intelligence into autonomous cars.

Mr. Eichhorn, the Volkswagen Group wants to make progress on electric mobility, autonomous driving and connectivity. At the same time, development spending is supposed to decline. What does this mean for R&D in concrete terms?

Ulrich Eichhorn: We consider the major issues to be productivity and efficiency on one hand, and innovation and digitization on the other. We have explicitly set the goal of becoming even more efficient in our traditional disciplines to free up more resources for innovations.

Boosting productivity and efficiency to finance new areas. Will this equation work?

It is naturally easier said than done. But there are solid connecting points. For example, we are reworking our entire development process, starting with our software tools and systematic reductions in complexity, all the way to a stronger development network to link the brands, to more fully activate synergies. We are also promoting the virtual development of new vehicle models and quickly moving engineers into new fields if they are losing their relevance in the new world of mobility. This will make us even more agile. Thanks to many individual measures, I am convinced that we will be in a position to strengthen our strategically important research and development areas.

Lawmakers’ ambitious CO2 goals require a rapid transition to electric mobility. What will the company’s next steps in battery technology look like?

To meet our CO2 goals, we are betting on a purely electric battery powertrain as well as plug-in hybrids. By 2025, we want to bring more than 30 new electric models to the market. Right now, we are seeing that plug-in hybrid vehicles are very popular with our customers, in part because concerns about the range of purely electric powertrains resonate with many drivers. But that will change quickly. We have already announced ranges of up to 600 km for the Volkswagen I.D., our compact concept car. This finally solves the problem, especially when our own fast-charging infrastructure is installed. But with regard to the battery: This is naturally the crucial factor for the electric car, as is the cylinder head or the mixture control for the internal combustion engine. We want to understand all the details of this technology in sufficient depth that we can develop and manufacture everything ourselves.

You have talked about 30 new electric models. Is Volkswagen turning to conversions or purpose-built designs?

Both, though I am not a fan of the term “conversion design.” To me, it sounds a bit like a mishmash. From the beginning, we developed the MQB, our modular transverse matrix, in such a way that it can handle a range of different powertrains effectively, whether they are gasoline or diesel engines, plug-in hybrids or electric motors. I don’t consider it to be a conversion design. And last year, we exhibited the I.D. at the Paris Motor Show, a model that is based on the MEB and thus is designed to exclusively house an electric motor, and not any other type of powertrain. The MEB is our battery-based electric platform for the future.

Level 4 autonomous functions are expected to be a component of the I.D. models starting in 2020. Is that realistic?

Our Volkswagen I.D. is naturally a concept car, meaning that it is a vision. It would be a mistake to infer that all the functions moving in this direction will be available right at the start.

Has sensor technology already developed sufficiently to be integrated into the series production of new models?

Of course. You are familiar with the laser scanners on our concept car, the five units on the roof. And then there are stereo cameras and all kinds of sensors that are invisible from the exterior. We are betting heavily on sensor fusion and on a highly reliable environmental model in the background. Our vehicles combine different sensor data and map information into a precise depiction of the environment.

Audi is currently working on a central driver assistance control unit, the zFAS, which is expected to debut in the next A8. Will the system be rolled out across other brands?

Systems of this kind are basically developed for platforms. In its specific form, the zFAS was developed especially for premium vehicles based on the MLB. But we have an alternative sister development as an option for other models within the group’s brands. Levels 1 and 2 driver assistance in automated driving and as an enabler for other systems ”“ that is something that we will be establishing very broadly over the next few years.

Isn’t the purchase of such a solution from a supplier, from Nvidia or NXP, worth considering?

The central driving system, whether it is driver assistance or autonomous driving, is of such great importance that we mainly want to control it ourselves. The reason is that we will incorporate these systems into millions of vehicles. They are things that we must understand and master. Within these architectures, there are naturally partnerships. We are already working intensively with Nvidia. In each case, we use the best component technology available on the market. With this approach, the entire system is subject to our control.

Is artificial intelligence essential for Level 5 autonomous driving?

The first question could be: Is it even permissible? Let’s take self-learning systems, for example: We would never permit one of our vehicles to be driven autonomously with learning content that we have not previously approved. Consider how a number of traffic participants behave at an intersection, and suppose our vehicle treats the situation almost as a model and mimics them. No, that’s not our goal. An approval is always required for the actual driving function.

Nonetheless, there are cases in which self-learning systems can reasonably be used. For example, sensors that independently adjust the sensitivity of the lighting and improve over time. Something like this makes sense and is useful, which is why we are offering it to our customers. But we will never expose highly safety-critical systems to open-ended self-learning.

VW Group CIO Martin Hofmann is passionate about artificial intelligence. Are there points of contact?

Of course. We work together closely. Here is one concrete example of valuable cooperation: Previously, when we built a car in Wolfsburg, we never saw it again after it made it through the so-called reporting point 8 and the testing had been completed. Today, there is a digital depiction in our backend, and we can help customers from this backend by assisting repair shops in their work, for example. That only happens through close cooperation with corporate IT.

Is data sharing with partners already underway?

Yes, it is already happening today to a small degree. But there is still no business model behind it. For example, the discount that you can realize under the insurance law is so small that it is hardly worth the expense. But perhaps that will change very soon. Then, if customers want their vehicle and personal data passed on to their insurer, we could enable that technically.

Volkswagen has made the decision to build its own digital platform. It is supposed to generate 1 billion euros. Is this sum realistic?

The Volkswagen Group generates more than 200 billion euros in revenue a year. It sounds to me as though one-half percent more is definitely feasible.

In Moia, Volkswagen has established its own brand for mobility services. How do you envision such an ecosystem?

It is still too early to outline it fully. I can basically say that Moia services ultimately belong to shared autonomy, not owned autonomy. The services blend seamlessly with electric mobility and autonomous driving. In corporate research, we have been working for a fairly long time on mobility on demand and dynamic routing. But it is only now that the IT landscape has reached the point that the services are becoming really attractive to customers. Previously, you would have had to call somewhere and a dispatcher would have assigned a car to you. In the age of the smartphone, this works much more smoothly. In this case, you just tell an app: Pick me up and bring me to the airport. And I receive a confirmation immediately that says “Be out at your door in three minutes.” That is something totally different.

A look ahead: Don’t the modular platforms still need more tightening to head into the future efficiently?

Our platform strategy is the foundation of our success as the world’s largest automaker. We are reaping what we sowed with the MQB and MLB years ago. A drastic tightening of variability was a huge feat that is now bearing fruit. Certainly, the ease with which we can position new variations is occasionally a temptation. We are striving for the right balance between an appealing range of variations and exuberant complexity.

Finally, is the diesel scandal slowing the speed of the transformation that the Volkswagen Group is now undertaking?

To be honest, we would naturally have taken the money that we have had to spend elsewhere and gladly invested it in all the areas of innovation that we have talked about. But I can assure you that we are not skimping on the projects that are relevant to the group’s future.