Robot wars
BMW lays bare its plans for an autonomous future, and the harsh realities of tech development
“THE WHEELS are primarily there to keep the computers from dragging on the ground.”
So wrote Paul Saffo, US futurist and academic on autonomous cars, for Wired, in 2012.
Humans who can drive are brilliant. You, me, and all the others who do it well are performing a task of great complexity with apparent ease, at least most of the time. Exactly how skilled we are becomes clear when you examine in detail what it will take to put a fully autonomous car – one able to do it all without a driver – into showrooms.
That’s exactly where BMW is headed. The Bavarian brand has begun work on a car, currently called the inext and scheduled for launch in 2021. “It goes without saying that the inext will be electric,” says research and development chief Klaus Fröhlich. “But rather than the drivetrain – which will have a real driving range of 500km – the special thing about it will be that it’s our first car with fully autonomous driving capabilities.”
Note that last word, “capabilities”. It means BMW isn’t entirely confident laws allowing self-driving cars will be ready everywhere in the world when the inext launches. But the company is poised to make the giant leap beyond Level 2 (see news feature on p16) to Level 3, which it labels Highly Automated Driving, onto to Level 4, and Level 5 and beyond.
To convince the world it’s deadly serious, BMW recently staged an in-depth briefing near Munich on its planned path to self-driving car production. The advanced self-driving technologies BMW has been developing for a decade will soon make the move out of prototypes and into cars rolling down an assembly line. Or as Peter Varadi, a senior engineer working on the program puts it: “Now we are in the industrialisation phase.”
BMW doesn’t hide its ambition to be the leader in autonomous driving. According to a recent survey by Navigant Research, it is among the front-runners in the race to bring fully autonomous tech to market. The Chicagobased company ranked BMW No. 6, ahead of outfits with much higher profiles in the field, like Google spin-off Waymo, Uber and Tesla. And all five of the brands ranked ahead of BMW by Navigant Research are existing car makers; Volkswagen Group, Daimler, the Renault-nissan Alliance, GM and, at the head of the pack, Ford.
What sets BMW apart from the others is its willingness to discuss how it plans to go about the fouryear job of designing, engineering and developing a self-driving car for mass production.
The Bavarian brand won’t go it alone. It’s too big a task. “We want as broad a base as possible for our collaborations and are open for new partners from the automotive, tech and mobile communications industries,” says R&D boss Fröhlich. BMW will adopt an open and co-operative approach that supports the development of industry-wide standards along the way.
BMW has already partnered with US computing giant Intel, which recently acquired the Israeli camera sensor specialist Mobileye. And along with Audi and Daimler, BMW owns a share
of high-definition digital map company Here. In May, major component supplier Delphi joined the grouping. It plans to make the tech developed first for the inext available to all.
For its part, BMW will massively boost the number of people it has working on autonomous driving tech, from around 600 at the moment to 2000 in the near future. It’s building a special facility near Munich for them to work in. And, through the second half of this year, it will complete the build of a fleet of 40 7 Series fitted with prototype autonomous tech.
Technology able to duplicate what we drivers do with our eyes, ears, hands, feet and brains isn’t going to be cheap. “There’s a lot of development cost initially,” admits R&D chief Fröhlich. “The first autonomously driving car is not going to make us any money, that’s for sure.”
Why does BMW believe that now is the time to start work on their entrant in the selfdriving car race? Because tech advances make it commercially feasible, the company’s experts say. While better sensors, connectivity, machine learning and artificial intelligence are all crucial ingredients, the imminent availability of next-gen highperformance computing seems to be the most important.
Futurist Paul Saffo is wrong about the autonomous car. The wheels will be there to do what they’ve always done; to provide a basis for personal mobility. The computers will be there primarily to make the wheels go where they should, while bringing massive reductions in road accidents, time wasted behind the wheel, energy consumption and congestion. All of which BMW promises its inext will deliver…
“The first autonomously driving car is not going to make us any money, that’s for sure” Klaus Fröhlich, BMW research & development chief