Today 1.2 million people die every year in car accidents. Ninety per cent of them are caused by human error. That’s 25 people a second, 150 people since you started reading this. We need to save ourselves from ourselves. Robots to the rescue…
The future of the car is in the hands of the suppliers inventing tech for a bunch of carmakers steeped in methods of manufacture that haven’t changed much since Henry Ford came up with the assembly line. The car of tomorrow will still come with a carmaker’s badge on the bonnet, or whatever you’ll call the robot’s front bit, but really carmakers will become a facilitator of tech from companies like Google, Nvidia, Qualcomm, camera makers, radar makers, laser makers, and Bosch.
Stephan Stass heads the German supplier’s automated technologies department and spoke to wheels about the future of artificial intelligence and electric vehicles at the Bosch proving ground in Germany.
If we look at the history of the automobile, our human environment — streets, highways, cities, suburbia — has been shaped by the car. How will future cars change the human habitat?
Stephan Stass: Well, I think, I lived in megacities for a couple of years, and I can already feel that the behaviour of users and the demand for mobility is changing. So young people, they are not even any more interested in getting a driver’s licence. So they simply have a demand for going from A to B. This is simply a mobility demand. It’s not like, “I must own a really nice car, with a lot of horsepower.’”
In megacities I think this is changing — in megacities cars which only fulfill this kind of mobility demand, they need to be autonomous, right? You want to have them when you click on your smartphone, and they should come, and they should bring you where you want. And you should not drive, or you don’t want to drive…
Is there a conflict between this want, this autonomous need, and people not trustful of this technology at the same time?
SS: What we see especially with these first steps of autonomous driving… it’s all about us getting used to this autonomous driving stuff.
Obviously we have to start at lower speeds — first of all in these kinds of traffic jam situations. Everybody says “I want that, I want this function, this is really great.” Nobody wants to drive in a traffic jam, nobody. If lots of people have this function available, and can experience this function, this will increase the perception of reliability of this system.
Look, in the beginning, even if you remember back when this ‘ESP thing’ came. Everybody though, “Pah, I don’t need ESP. I can drive.” You know?
It all sounds very utilitarian… With less and less ownership desire and more shared mobility, are established car companies challenged with this erosion of their brand desirability?
SS: I would say… Well, I don’t know. Simply, the mobility demand is changing. And I don’t think that can hardly be influenced at all, because it is influenced by lots of other things. Young people, they want something completely different — I have a nice car, and I think, “Hmm, nice car…”, but my little daughter is not interested in that. She is not interested in nice cars — she is interested in smartphones. So this is simply shifting.
But there is a big discussion going — will this completely destroy the volume of the automotive industry? This is a scary thing, but so far, people are not that scared of it.
For example, you have a shared autonomous vehicle, you know, it’s running all the time. Today’s cars are parked all the time… They are not mobility solutions, they are parking solutions. They stand still for 23 hours of the day. And future shared cars will drive all the time. This will have a completely different level of wear, so maybe after three years you need a new car, while today’s cars can run much longer.
Isn’t that the opposite of sustainability? Cars that have much shorter usable lives?
SS: The question is how far a car can drive, not how long it can drive.
We all talk a lot about autonomous driving, but not so much about autonomous parking. If we look at megacities of today, 30 per cent of urban traffic is caused by parking issues. So you have CO2 problems and all this fuel wasted.
Many problems in traffic are only caused by people looking for a parking place. So if you remove that, you also answer a question from an environmental standpoint.
In layman terms, Artificial Intelligence — where are we at the moment, and how does it compare to the human brain in terms of, well, processing power?
So, the latest technology, from a performance points of view the GPUs are in the range of 30 terraflops per second. This is much more than a human brain can handle. But this is not a fair comparison.
Because, you know, the question is how to train AI, and what is its learning capability? And there I would say we are making fantastic progress, but we are still at the beginning.
We can already train AI behaviour, and interaction between drivers quite well, but the capability of just a small child is really bigger. A child, with you walking and holding on behind, after two minutes can learn to ride a bicycle… I would say that this is still a real challenge. This kind of human learning is still excellent, but on the other hand human beings still have these kinds of deficiencies I would say, where for example in an emergency situation you only start to react after one second. Our computers don’t need a second, they need a microsecond, they can react immediately. And you know exactly just how far a car can go in one second. So this already saves lots of lives… We expect robo-cars to drive better than human beings in the next decade. This will happen.
If Bosch has the technologies now, the e-axle that is basically one box that you attach wheels to and you have a drivetrain, doesn’t this disrupt the traditional carmakers? Deutsche Post bought a bunch and made their own delivery van, instead of going to VW to buy a fleet of new vans. Aren’t you in a way taking bread off the carmakers’ table?
SS: Not more or less than before. We don’t build cars ourselves, and this is not our intention. Let’s say we have all the technology available to build cars, but it’s not our intention of to go into any competition with carmakers or with startups.
But if start-ups or smaller companies come to us, and want our technology, for us this is another valuable customer. For sure we love to work with the big ones, but from out point of view it’s another customer.
There’s more work to be done outside of the vehicle than inside the autonomous car of tomorrow, this world filled with smart roads and smart cities… Who’s going to build this world?
SS: Very good question… So, first of all, in the case of autonomous driving and recent developments we are not looking for an infrastructure. We are simply looking for an internet connection. But from the year 2019, 2020, and onwards, in the developed countries every car will have an internet connection, with or without autonomous driving capability. So because of this fully-connected environment, this is a vehicle feature. So if you ask me who is paying for this, it’s the customer. It’s a feature. As everything is connected and the car is then the internet of things, services which bring value to customers, they will pay for. Maybe we cannot yet imagine even the services that will be available. For example HD maps are a completely new thing, and a completely new service that lets you localise your car so precisely. So now a normal market process starts — you know, there is a customer, there is demand, there is competition, and then prices are market-determined.
As these things become apparent, it will just be competition in the marketplace breeding development?
SS: Maybe there will be some shift in values here. There will be completely new players in the marketplace, taking much more care about service offerings in driverless cars.
Projections show about 20 percent of cars will be AVs or EVs by 2030 - can you see a coming day, one day a long while from now, when the human-controlled cars are outnumbered by AVs. When they are ‘shunned’ from society, banned from roads occupied by AI?
SS: I can imagine that something like this could happen. Maybe starting with megacities, where driving will be limited. I am convinced of that because there is one fundamental thing, and this is avoiding accidents. Currently over one million people die in road accidents every year. So, this is an unbelievable number of people dying…
I think if we can eliminate this by automatic driving, then…
What about the responsibility factor when it comes to AI?
SS: We have quite a lot of experience in such products. I’m not talking about autonomous driving now — if we look back at this ESP thing. We pioneered this technology a number of years ago and now it’s mandatory in almost all countries, that ESP must be in the car.
But there was a lot of discussion in the begging, you know? At that time, we thought, maybe this is even more dangerous? To have something like ESP, because some drivers will now drive even more risky, because they think they have a guardian angel with them in the car at all times.
But actually, this didn’t happen. This didn’t happen at all.Now every car has ESP and most people don’t even know they have it. But what has happened since ESP was introduced? In Germany alone, one third of all fatal accidents have been eliminated. One third… No government, and no society can simply ignore this kind of potential. That is why I am convinced that automated driving will come.