Autonomous driving Johann Jungwirth: "The main task is to test, test, test"

Related Vendors

FAULHABER Drive Systems

Wuxi InfiMotion Technology Co., Ltd.

Johann Jungwirth is convinced of the future of autonomous driving. Comfort and especially safety for users are crucial to him.

Johann Jungwirth, Senior Vice President Autonomous Vehicles at Mobileye, enjoys driving autonomously himself and is convinced of the safety and comfort. He believes that private customers will also be willing to pay the high price.

Mr. Jungwirth, with the Drive Pilot by Mercedes-Benz there is already a Level-3 system on the market. However, these are fair-weather pilots that only work during the day in good visibility up to 60 km/h. The feature costs around 7,000 Euros. Do you see a willingness in the market to pay this price?

From our perspective, it is important to reduce limitations and also make speeds up to 130 km/h possible. This means that the operating range must be broad enough to also handle darkness and light rain. As a driver, one has to feel that the system actually always works and that one only needs to intervene in exceptional situations. These are the solutions we are working on.

What makes the case for the assistants, more the argument of comfort or safety?

It's both. Safety is a basic requirement; without it, no approval from regulatory authorities is received. If you drive 10 times or 100 times safer than a human, then it's a strong argument to activate the system. Being able to use time differently is an aspect of comfort: reading, working, learning – even before level 4.

The price of the systems suggests that level 4 functions will mainly be used in commercial vehicles, or is this assumption incorrect?

No, probably not at the moment. We are witnessing the development towards the robotaxi, for example with the VW ID Buzz AD at Moia. They already face the challenge of not finding enough drivers today. To offer such a service, you need a network effect, meaning a certain density of vehicles on the road, so that customers do not have to wait long. Also, spontaneously reacting to changes in demand is only possible with autonomously driving vehicles.

Moia aims to be autonomous in Hamburg by 2025. Do you consider this timeframe realistic?

Yes, I consider that realistic. In Oslo, our partners Holo and Ruter are pursuing a similar goal. Norway was also a leader in the introduction of electromobility, thus creating the right framework conditions. Oslo is planning with up to 30,000 autonomously driving vehicles, three times more than in Hamburg. That's how you get the desired network effects. For users, this primarily means: no long waiting times.

Do you really believe that a private customer will buy a level 4 car? The price will be significantly higher.

Definitely. If there will be the option to simply enter the destination and the car drives you there independently, then this can be appealing for some users. Even today, there are people who afford such a luxury – just with a human chauffeur instead.

I was thinking more of average earners as the target group. The car is likely to be the most expensive investment after the house and is usually only used for a short period of time.

We will see. That's why we at Mobileye are very broadly positioned and say: the market will regulate this. Whether it will be fleet vehicles or rather private vehicles in the long term, nobody knows today.

At this year's CES, the Software Defined Vehicle was the common thread among exhibitors. Car manufacturers must become masters of every line of code in the car. Can the established manufacturers do this?

That's why we introduced our Driving Experience platform at CES 2024. The idea: We develop all essential driving functions universally. Everything that does not differentiate from the competition is used by all OEMs together. This includes, for example, the environmental detection of the sensors. It has to work perfectly.

Everything that is unique to the manufacturer can be customized on our platform. OEMs thus have access to the application layer, libraries, and tools. In principle, this allows the driving style of a vehicle to be completely individualized. Every car must stop at a red light, but how exactly the vehicle approaches and performs the braking process is up to the manufacturers. In the end, the car drives as if the manufacturer had programmed the entire system themselves. We see this as the golden mean between an open and a closed system. After all, one does not need to reprogram iOS from scratch for a phone app each time.

How do you see the public acceptance of robotaxis? Has it suffered due to the license revocation for Cruise in the US following an accident involving an injured woman?

The entire industry above all must draw the right conclusions from this and continue to work focused. Today, 94 percent of all accidents are caused by human error, namely inattention, fatigue, alcohol, or drugs at the wheel. With robotaxis, all these reasons for accidents are eliminated. What is essential is that no new reasons are added due to the technology. Ultimately, it is a matter of testing. Therefore, the main task for the coming years is: testing, testing, testing. Especially those rare and unexpected cases, which we call corner cases.

This brings us to the area where it can be proven that the machine drives safer than a human. First, it's factor two, then ten, and eventually factor 100. In addition, the reaction speed of the systems is much faster than that of humans. The machine reacts within 100 to 200 milliseconds, while a human has a moment of shock, needing 800 milliseconds to a second to properly assess the situation and then react. And the sensors also see well at night. We rely on triple redundancy in our sensor set: cameras, radar, and lidar.

Are you sticking with the trio of sensors, meaning camera, lidar, and radar?

As of today, we are starting with cameras, lidar, and radar. So, imaging radar sensors, for example, can be seen on the second generation of the ID Buzz AD. We offer 1,536 virtual channels in the front imaging radar sensors and 384 virtual channels on the sensors at the sides and in the rear area. This means we have a radar with a resolution similar to a lidar sensor. At the same time, radar technology is more robust and less expensive.

What do you think of thermal imaging cameras that very reliably detect people and animals on the road?

Sensor technology will continue to evolve; there will be many new things, new frequency ranges, black light, and so on. These will all be improvements and further developments of the existing sensor technology. We are very focused on making the cost factor of the systems attractive – also with regard to level 4 for private users.

When you drive, what has been the longest distance so far that you have allowed the assistant to drive with a clear conscience?

The longest distance must have been the two and a half hours from Jerusalem to Haifa via Tel Aviv. I live with my family in Israel. On weekends, we often drive to the Dead Sea and back, which I drive almost completely autonomously.

Are we talking about a level 2 plus system or higher?

I have to try out all of our products, from Supervision to Chauffeur to Drive. I really enjoy experiencing it for myself. I am a combination of a technology and car guy.