The technology required for self-driving cars is proving difficult to perfect. The further technology advances, the more scientists realize that the current generation of computers cannot overcome the last hurdles.
The car is, from a safety point of view, a disaster on wheels. If two Boeings full of passengers were to crash every year in the Netherlands, no plane would go up in the air, but we all take it for granted that just as many people are killed in traffic every year. Every day 3500 people worldwide die in traffic and cars are the number one cause of death among young people in the US.
Significant steps were taken in the 1960s to reduce the number of road deaths. Think of things like three-point seat belts, crumple zones and laminated glass. In the decades that followed, the introduction of airbags, ABS (anti-lock braking system) and ESP (traction control) saved many lives. But unfortunately, drivers haven’t improved or slowed down much since the 1960s, so it makes sense that self-driving cars are seen as the way to reduce the number of deaths.
Elon Musk
In theory, it’s also a fantastic plan. After all, a self-driving car does not drink, never gets tired, is always alert and never crashes. And so there was great optimism when Tesla CEO Elon Musk said back in 2014 that Tesla’s will be fully self-driving ‘next year’. It turned out to be a promise he would repeat for nearly eight years in a row. Since 2013, other automakers have also claimed that fully autonomous cars would be in showrooms by the end of this decade. New cars were sometimes even shown photos of versions without steering wheel. Until a few years ago, there was almost daily news about the development of self-driving cars.
And indeed, there are already cars that can drive independently on the highway: accelerating, braking, driving in traffic jams and staying within the lane until a certain speed. Teslas can already overtake independently and companies such as Waymo (a subsidiary of Google), Uber and Cruise (a subsidiary of General Motors) have been testing prototypes of completely self-driving cars on public roads for years. In San Francisco you can order a robot taxi without a driver. But the further technology advances, the more uncertain the sounds from the car industry.
No trust
The number of incidents involving self-driving cars has grown steadily and nowadays you hear little about it in the media – just like the flying car. Unless something has gone wrong again, because drivers don’t trust self-driving cars. A JD Power survey in Germany found that nearly two-thirds of those surveyed do not trust self-driving cars.
“The ultimate goal is to make cars that are safer than human-powered vehicles,” said Philip Koopman, a computer professor at Carnegie Mellon University. In the US, there is about one death for every 100 million miles driven by a human. Koopman says AV makers need to beat this figure to prove their technology is safer than a human. “Safety has everything to do with the rare case that it does not work properly,” says Koopman. It should work 99.99999999999 percent of the time. Automakers are still working on the first few nines, with a lot of nines to go. For every nine it is 10 times harder to achieve that.”
Gabriel Seiberth, CEO of the German computer company Accenture, also thinks that computer technology can never become so good that it can match the human brain, not even with artificial intelligence. “A feasibility study by our company has shown that this is not possible. One of the core problems of artificial intelligence is that systems can only be trained for a special task. If it goes beyond that task, the computer won’t know. Their learning behavior is not based on the experimental behavior of children, but through adaptation based on examples. I also don’t think it’s smart that car manufacturers focus on this. They better try to make money with what they can.”
robot taxi
Carlo van de Weijer, director of Artificial Intelligence at Eindhoven University of Technology, and a specialist in the field of smart mobility, does think, unlike Seiberth, that computer technology will one day be as smart as humans. Yet he also sees little future for fully self-driving cars. “Cars are going to offer more and more technical support to increase safety and partly autonomous driving is part of this. Then you can do something else on the highway in traffic jams or during your rides on boring, long stretches. I also believe in it as a shuttle to and from parking lots and at attractions, but a car that completely takes over all tasks from us will not come.”
In his column in Het Financieele Dagblad, Van de Weijer already wrote about ‘The myth of the robot taxi’. In it he announced that Elon Musk wrongly enthuses people to buy a Tesla, because then you could earn money by using your car – if you don’t drive it yourself – as a robot taxi. ,,That’s never going to happen. Not only because it is technologically impossible, but also because there is no demand for it. The only thing self-driving cars lead to – if they ever happen – is a bad kind of public transport. Governments will never accept that and neither will motorists. They are too attached to their own car for that.”
Free unlimited access to Showbytes? Which can!
Log in or create an account and don’t miss out on any of the stars.
#Selfdriving #car #threatens #fail #wont