I think for the passenger cars, the value will be the strongest in these big megacities.
March 15, 2022
When it comes to autonomous vehicles, safety is the leading concern on people's minds. The early cases testing this technology in urban robotaxis and driver-less buses in less populated areas is helping identify safety issues and cost aspects that must be addressed before autonomy is introduced.
In this second part of three interviews, Ralf Klädtke, TE's vice president and chief technical officer for Transportation Solutions, explores the human side of autonomous driving, including trust issues with AI. Klaedtke saw the early adoption of this technology.
Read – and listen to – an interview with Ralf Klädtke.
Where will autonomous vehicles first be used - cities, countryside, or suburbia?
I think we again have to talk about the passenger cars and the commercial vehicles. I think for the passenger cars, the value will be the strongest in these big megacities. If we can block the inner circle and can have robotaxis going in and out, because then you have shared mobility, a lot of pleasure of operating, and that's a very strong value. The next value on the passenger car and as well on the trucks I see on the highways, because here we have usually well-defined areas. So, the autonomous trucks, the freight networks will come first. That's a strong value. And the passenger cars with the highway pilot will follow.
And here, I think the only barrier is still the liability. Whether the liability stays with the driver or the liability costs to the car maker, I think that is the key. And I think after the megacities or some restricted areas, if you look to, for example, inner cities, shuttle buses going at slow speed, people trust that people go into a shuttle bus going at slow speed because people feel safe. And I think the next level of application and value will fully follow on the countryside. I think that's the next level to come. Mixed traffic, high speed, different weather conditions. I think this is a very long way until we feel safe in this area and it's a cost aspect as well.
How will humans learn to accept vehicle autonomy?
I think if you look across the world, it's very different where you are. We see an openness for technology in the Asian Pacific group, where people adapt much faster to new technologies, are more open to that. But people in general, I think 70% of the people in survey, say that technology is making the world a better place. So that's one point. But when it comes to autonomous driving, then still more than 60% don't trust it. So we like technology. We don't trust it yet. And therefore, it's important that when we enter into the autonomy game, that safety is coming first and we go step by step. So, if you talk to someone, if you have an autonomous shuttle at an airport, people will enter into that, trust it to drive you to the next terminal. No problem. If we have a restricted area, a dedicated autonomous line, people will trust it. I think it's up to us to create an environment where autonomy works, where we can make sure that it's safely operated, and then autonomy can be rolled out.
Is artificial intelligence smart enough to give people people confidence in autonomous vehicles?
So I tried to talk to the artificial intelligence. If you want to have an intelligent, educated discussion with an AI, there will be a lot of surprises. I think if you compare the IQ of an AI right now with a piece of toast, it's about the same level.
And artificial intelligence is a machine learning device. So, it's really great in analysis of images, of pictures, of objects, of calculating distances, and to try to have high processing times in order to make sure that we can operate safely. So, it's great at that. But it has to be trained for all kinds of scenarios. And the human brain is really capable of scenarios that you don't know before. So some things really, if you operate this shuttle in the inner city and you have only one lane and the autonomous shuttle is operating on that lane, then the car is parked in order to just shop some newspaper. Someone parks it in the second row, it gets out, wants to have a newspaper. And then there is a white line. So anyone of a human driver would cross the white line to get around that car. An autonomous vehicle would just stop and stand still. It will never break the rule. It will be safe. So, these are scenarios that you cannot program for things where the human brain says, okay, there is no risk to pass that car. I might violate this rule, but I will operate safely. This tremendous amount of scenarios you cannot train for, and that is something that will move ahead. The artificial intelligence will get better, the cars will get trained, and we get on a safe route on that.
How might vehicle autonomy be introduced at scale?
I think if we have these megacities, inner circle, restricted areas, controlled lanes, we will say this is for autonomous vehicles only. This will get all of us a big step forward. And another point which will support that is the V2X vehicle to everything when more and more will be 5G communication between the vehicles, the infrastructure, everything will communicate. Then you will see more and more that every vehicle can tell, especially the autonomous ones, my path will go along here. Another car will already know at what speed, what kind of course you're going to take before it's even seen, before any sensor can see that. The flow of traffic, especially when you have more and more autonomous traffic, will continuously improve. But I think the baby steps are the right approach for all that create restricted areas for autonomous traffic. Don't mix traffic whenever you can and get into controlled good weather situations to provide a safe feeling for all humans that autonomous driving is getting now at the level of maturity that we can operate safely and that we are on the mission to zero fatalities.
Do you believe that increased autonomy will eventually lead to better safety on our roadways and that mission to zero fatalities?
I think if we look, where are we today? Today we have globally 1.3 million people killed in traffic accidents every year. To me, that's a totally unacceptable number. So, governments signed up for the mission to zero fatalities. And the autonomy we talked about going to level two plus for the passenger cars provides additional sensors, additional visibility. The human driver will be supported to keep in the lane, to have emergency braking, to see objects that the human eye might overlook. Or when you are tired, sleepy in situations like that, that there is additional safety measures. So, this will already on the passenger car get a lot of risk out of the system. Same with the truck drivers in cities. When a truck gets around the corner and someone on a bicycle is just in the area that is not visible for the driver. Sensors support and sensors can react, can initiate an emergency braking can save a lot of lives. Therefore, the autonomous driving technology, even though it's a longer way to get to autonomy, leads us on the path to zero fatalities as much as possible. So, it's a big, big differentiator to save a lot of lives in traffic.
Did you enjoy this interview? Read the source article.
In the race to achieve vehicle autonomy, it is not important which type – passenger cars, fleet trucks, or robotaxis – win. What matters is that we as an industry take a holistic approach to developing autonomy that addresses sustainability and roadway safety with zero fatalities.
Achieving level-5 autonomy in cars means addressing challenges not initially envisioned. Learn about these.
I think for the passenger cars, the value will be the strongest in these big megacities.
March 15, 2022
When it comes to autonomous vehicles, safety is the leading concern on people's minds. The early cases testing this technology in urban robotaxis and driver-less buses in less populated areas is helping identify safety issues and cost aspects that must be addressed before autonomy is introduced.
In this second part of three interviews, Ralf Klädtke, TE's vice president and chief technical officer for Transportation Solutions, explores the human side of autonomous driving, including trust issues with AI. Klaedtke saw the early adoption of this technology.
Read – and listen to – an interview with Ralf Klädtke.
Where will autonomous vehicles first be used - cities, countryside, or suburbia?
I think we again have to talk about the passenger cars and the commercial vehicles. I think for the passenger cars, the value will be the strongest in these big megacities. If we can block the inner circle and can have robotaxis going in and out, because then you have shared mobility, a lot of pleasure of operating, and that's a very strong value. The next value on the passenger car and as well on the trucks I see on the highways, because here we have usually well-defined areas. So, the autonomous trucks, the freight networks will come first. That's a strong value. And the passenger cars with the highway pilot will follow.
And here, I think the only barrier is still the liability. Whether the liability stays with the driver or the liability costs to the car maker, I think that is the key. And I think after the megacities or some restricted areas, if you look to, for example, inner cities, shuttle buses going at slow speed, people trust that people go into a shuttle bus going at slow speed because people feel safe. And I think the next level of application and value will fully follow on the countryside. I think that's the next level to come. Mixed traffic, high speed, different weather conditions. I think this is a very long way until we feel safe in this area and it's a cost aspect as well.
How will humans learn to accept vehicle autonomy?
I think if you look across the world, it's very different where you are. We see an openness for technology in the Asian Pacific group, where people adapt much faster to new technologies, are more open to that. But people in general, I think 70% of the people in survey, say that technology is making the world a better place. So that's one point. But when it comes to autonomous driving, then still more than 60% don't trust it. So we like technology. We don't trust it yet. And therefore, it's important that when we enter into the autonomy game, that safety is coming first and we go step by step. So, if you talk to someone, if you have an autonomous shuttle at an airport, people will enter into that, trust it to drive you to the next terminal. No problem. If we have a restricted area, a dedicated autonomous line, people will trust it. I think it's up to us to create an environment where autonomy works, where we can make sure that it's safely operated, and then autonomy can be rolled out.
Is artificial intelligence smart enough to give people people confidence in autonomous vehicles?
So I tried to talk to the artificial intelligence. If you want to have an intelligent, educated discussion with an AI, there will be a lot of surprises. I think if you compare the IQ of an AI right now with a piece of toast, it's about the same level.
And artificial intelligence is a machine learning device. So, it's really great in analysis of images, of pictures, of objects, of calculating distances, and to try to have high processing times in order to make sure that we can operate safely. So, it's great at that. But it has to be trained for all kinds of scenarios. And the human brain is really capable of scenarios that you don't know before. So some things really, if you operate this shuttle in the inner city and you have only one lane and the autonomous shuttle is operating on that lane, then the car is parked in order to just shop some newspaper. Someone parks it in the second row, it gets out, wants to have a newspaper. And then there is a white line. So anyone of a human driver would cross the white line to get around that car. An autonomous vehicle would just stop and stand still. It will never break the rule. It will be safe. So, these are scenarios that you cannot program for things where the human brain says, okay, there is no risk to pass that car. I might violate this rule, but I will operate safely. This tremendous amount of scenarios you cannot train for, and that is something that will move ahead. The artificial intelligence will get better, the cars will get trained, and we get on a safe route on that.
How might vehicle autonomy be introduced at scale?
I think if we have these megacities, inner circle, restricted areas, controlled lanes, we will say this is for autonomous vehicles only. This will get all of us a big step forward. And another point which will support that is the V2X vehicle to everything when more and more will be 5G communication between the vehicles, the infrastructure, everything will communicate. Then you will see more and more that every vehicle can tell, especially the autonomous ones, my path will go along here. Another car will already know at what speed, what kind of course you're going to take before it's even seen, before any sensor can see that. The flow of traffic, especially when you have more and more autonomous traffic, will continuously improve. But I think the baby steps are the right approach for all that create restricted areas for autonomous traffic. Don't mix traffic whenever you can and get into controlled good weather situations to provide a safe feeling for all humans that autonomous driving is getting now at the level of maturity that we can operate safely and that we are on the mission to zero fatalities.
Do you believe that increased autonomy will eventually lead to better safety on our roadways and that mission to zero fatalities?
I think if we look, where are we today? Today we have globally 1.3 million people killed in traffic accidents every year. To me, that's a totally unacceptable number. So, governments signed up for the mission to zero fatalities. And the autonomy we talked about going to level two plus for the passenger cars provides additional sensors, additional visibility. The human driver will be supported to keep in the lane, to have emergency braking, to see objects that the human eye might overlook. Or when you are tired, sleepy in situations like that, that there is additional safety measures. So, this will already on the passenger car get a lot of risk out of the system. Same with the truck drivers in cities. When a truck gets around the corner and someone on a bicycle is just in the area that is not visible for the driver. Sensors support and sensors can react, can initiate an emergency braking can save a lot of lives. Therefore, the autonomous driving technology, even though it's a longer way to get to autonomy, leads us on the path to zero fatalities as much as possible. So, it's a big, big differentiator to save a lot of lives in traffic.
Did you enjoy this interview? Read the source article.
In the race to achieve vehicle autonomy, it is not important which type – passenger cars, fleet trucks, or robotaxis – win. What matters is that we as an industry take a holistic approach to developing autonomy that addresses sustainability and roadway safety with zero fatalities.
Achieving level-5 autonomy in cars means addressing challenges not initially envisioned. Learn about these.