High cost of self-driving vehicle development could send some car-makers to the wall
Extreme development costs will eventually shrink the number of autonomous vehicle (AV) makers and start-ups into just a small handful of plug-and-play technology ‘stacks’, according to BMW’s development chief.
BMW’s Director of Development Frank Weber has also implied that AV development costs are so high that some car-makers may go to the wall if they insist on developing autonomous driving systems in-house.
While there are dozens of credible organisations working on assisted- and full-autonomous driving today, Weber believes the technologies will inevitably coalesce into just two or three.
BMW is slated to launch its first Level 3 car – the next-generation BMW 7 Series limousine – next year, and its iX electric SUV is fully-prepped for Level 3 self-driving capability, but that doesn’t mean the German premium brand is relying solely on its own technology for the future.
“What is obvious is that individual OEM autonomous solutions don’t make sense,” Weber admitted.
“There will be at the end a consolidated number of technology stacks because it’s very simple: many will not be able to afford that.
“By the way, since it’s such a regulated area, it does not even make sense to differentiate yourself,” he said.
The costs will be unworkable, Weber insisted, even for some very major corporations.
The car-makers leading the way in AV research and development are widely regarded as BMW, Audi, Mercedes-Benz’s parent Daimler, the Volkswagen Group, Ford, GM and Toyota.
Tesla is not considered a frontline player, based on its lack of registered AV test miles in its former home state of California, as well as relying on a camera-only hardware suite.
The PAVE campaign, a non-profit group started to inform the public about autonomous cars and how they could benefit daily life, has more than 80 members, and 25 of those are start-ups.
While he doesn’t suggest most of them will be wiped from the future world of autonomous technology, Weber believes many of them will coalesce, or at least form alliances.
He even teased that BMW may license out its own technologies (developed with Intel and Moblileye) for other OEMs and suppliers to use.
But far from abandoning BMW’s own research into autonomous driving, Weber insists its stacks are world leading.
“I have no interest in exclusivity. No interest at all,” he said (BMW welcomed Fiat Chrysler Automobiles into its AV research program in 2017 and Magna has joined since).
“For us, leading the industry is something, so when we have relations with suppliers they use our tech stacks.
“We have with this tech stack… there is so much software and money and knowledge in there that for some people that is very interesting as a starting point.
“We have developed this as a real product, and some people who are developing chips see this as an interesting platform and they help us to scale solutions where we can still lead and then have an interest to get it out into the industry.”
BMW itself has already headed down the alliance path, with almost 2000 workers in a dedicated autonomous-driving campus at Unterschleißheim, near Munich, in addition to offices for Mobileye, Intel and Delphi.
In Europe, the only major market with firm legislation in place for autonomous cars, Weber estimates at least 435 million miles of validation testing (700 million kilometres) are required to prove the software and hardware to the satisfaction of lawmakers, blowing costs through the roof.
Another side effect, he insisted, was that new models would have to launch without their Level 3 systems activated, as companies racked up the autonomous mileage behind the scenes to validate them.
“We will have more and more Level 2+ functions with relatively affordable technology with a very good level of assistant technology, with a driver in control and a good level of support whenever we launch a car, but Level 3 will always come later,” Weber said.
“When you make this huge step cost-wise and what the system is doing into Level 3, this will be much more expensive in the beginning and very restricted.”
Weber argues that commercial vehicles and long-haul transport will make the first major jumps into fully-autonomous production vehicles, for a number of reasons.
“It will be highways first, with separated lanes and the speed is restricted. The urban environment compared to a highway is a crazy situation,” he explained.
“Also, what you observe is that you can combine that with commercial vehicles.
“They go at lower speeds, always the same routes and they can start replacing drivers.
“They can live with restrictions [on routes], but in the normal car business we can’t.
“It will come, [but] it is a lot of effort and it takes more time than expected.”
Weber insisted that even the world’s biggest tech companies had underestimated the challenges of autonomous driving, with Google, Apple and Uber all going from leading players to bit players.
“Even Google. Autonomy is around the corner… It’s always around the corner… Google announced autonomy in 2012 and 2013 and it still does.”
BMW even tried an AV partnership with arch-rival Daimler in 2019 to speed up production to 2024, but it has been put on ice.
“We were in a partnership with Daimler and we stopped it.
“It runs counter to what I just said, but you find out that the timing is not aligning and you have almost this legacy of something that is already developed and the switchover costs with Daimler were wrong.
“We put it on hold [but] we both still agree and the co-operation still is very meaningful.”
Related: Driving will be illegal by 2050, says new studyRelated: What motorists want from self-driving cars
Related: Autonomous cars make other drivers feel less safe
Keyword: Most autonomous car programs won’t survive, says BMW