Computer Driven Autos Still Years Away Despite Massive Investment.
“Now if you want to have a L5 driving on a straight line through the middle of the desert, that’s much easier”
Despite the hype that liberation day will dawn soon and cars will drive themselves, there’s no realistic chance that full-on self-driving will be available before 2030, and then only in a tiny number of top-of-the-range sedans and SUVs, according to consultancy Accenture.
The world’s biggest auto manufacturers have poured billions of dollars into trying to perfect the technology. Despite impressive results, autonomous driving technology still can’t be relied on to handle really complex situations, and won’t any time soon.
Axel Schmidt, who leads Accenture’s automotive division globally, said in an interview autonomous vehicle (AV) technology had not advanced as quickly as expected.
“By now many expected that we’d see robotaxis driving around, but the problems are very complex. Driving on the highway when all the cars are going in the same direction is much easier than handling inner-city traffic. And much of the early research was done in California where there’s not much rain, fog or snow. In Europe’s all-weather conditions the problems were all a bit underestimated with problems like snow blocking road markings,” Schmidt said.
He said as the technology moved from level 1 and 2 competence, which covers simpler tasks like emergency braking, approaching traffic warnings, steering assistance and early connectivity, this still relied on the driver taking full responsibility for the car’s actions. Higher levels though mean that responsibility reverts to the manufacturer. Level 5 assumes total control by the computer with no steering responsibility for humans.
That currently is a bridge too far for the manufacturers, not to mention lawyers.
In a report, Accenture said many traditional carmakers, as well as new entrants, had taken a bold position and projected fully self-driving cars by 2025. That’s not going to happen, unless the technology can mimic human common sense.
“We expect that, by 2030, 60% of all new cars will be equipped with Level 2 features. The flip side is current systems are not robust enough to navigate cars autonomously in real-world traffic – arguably due to missing common sense. For example, an algorithm cannot distinguish between a real traffic sign and one that’s – for us humans – obviously manipulated,” the report said.
“Driven by high costs, we expect that even the share of premium cars with Level 3 or 4 will only account for about 5% of the total market by 2030. Mission impossible: Level 5,” it said.
Pedro Pacheco, senior research director at Gartner Group agrees.
“We won’t see L5 before 2030. Undoubtedly, some companies will try to generate some hype and claim they have it. However, having a L5 system that can drive from door to door in the whole of Europe and North America at least, for at least 80% of all itineraries and with 0 accidents and disengagements is something that we definitely won’t see this decade,” Pacheco said.
“Now if you want to have a L5 driving on a straight line through the middle of the desert, that’s much easier,” Pacheco added.
As well as the huge complexity of the autonomous driving task, major manufacturers have found other important issues which required additional capital and attention, including the production hiatus because of the coronavirus pandemic, sustainability, and government green deal requirements.
Accenture’s Schmidt said there are self-driving tasks which can be accomplished now. You could automate the movement of cars from production line to parking lot, or long-distance truck driving, parts of which could be easily automated.
But when autonomous driving is finally available, the industry faces some big decisions. Will the industry be driven by current manufacturers, mobility providers like UBER, new technology entrants like Apple, Google or Amazon or infrastructure players?
Schmidt said when Level 5 is achieved there will be big opportunities for businesses, driven by apps which can design travel plans to maximize mobility efficiency.
“With AV, we probably won’t see domestic flights as we see a seamless intermodular mobility system. For instance, I’d take a TGV (high speed train) from say Paris to Nice, drive to a town in the countryside in an AV. In the morning, I’d use an e scooter to get to the bakers. There would an app which would book these choices based on current traffic data, and my historic choice of travel, like Amazon today knows your buying habits,” he said.
This new world of mobility is up for grabs, according to Schmidt.
“There will be a big shift in how to handle mobility in the future. Who will be the coordinator
and owner, the auto manufacturers, big tech companies, infrastructure companies, fleet
operators, or maybe partnerships?”
Will any manufacturers fall by the wayside in this upheaval?
“The race is on and there will be new players like Apple. It will be a much harder race and
probably not all will survive. Breakthroughs are possible. Look at how Tesla has shaken up the traditional 100-year old industry,” Schmidt said.
These experts were interviewed ahead of the Mobile World Congress https://www.mwcbarcelona.com/ in Barcelona, Spain from February 29 through March 3.