In a Robot Car, the Car Is the Easy Part

If building a driverless vehicle were simply about putting parts together, we'd be living in the future already


General Motors has begun testing autonomous vehicles on public roads in Michigan. But it has a long way to go before it sells them to the public.

If building a driverless vehicle were simply about putting parts together, we'd be living in the future already

Here's what we know about autonomous vehicles: They'll be safer and smarter than the cars parked in our driveways today. Anyone who speculates beyond that is just, well, speculating.
 It wasn't long ago that a top automotive supplier was talking about providing fully driverless systems for production vehicles by 2019. Uber thought it would have 75,000 autonomous vehicles operating by this year. Elon Musk pointed to a similar horizon in 2014 when a group of Journal editors and reporters asked him when he thought an entirely autonomous Tesla would hit the road.
 If building a robocar were simply about figuring out how to best arrange a box of sophisticated parts, these predictions may have come true. Most of the necessary components— sensors, cameras, chips, those bulky lidar units that sit on top of the car—have been around for a while. Any car maker or parts supplier worth its salt could figure out how to gin up a remote-control SUV. But it takes gobs of engineers, data, software, patience and cash to teach that 4,000-pound vehicle to think for itself.
 “Our goal is not to build a car,” Dmitri Dolgov, the chief technology officer at Waymo, said at an MIT Technology Review event in March, shortly after the driverless-technology developer launched a robotaxi service in Phoenix that requires constant human supervision. “We're really building a driver.”
 Sacha Arnoud, one of Mr. Dolgov's colleagues, recently said it will take Waymo—which has already been working on building that driver for a decade—as much time to complete the home stretch as it took to get there. “When you're 90% done, you still have 90% to go,” Mr. Arnoud said during a separate MIT lecture. The first “90% of the technology takes 10% of the time.”
 General Motors Co. is one of the companies learning what it's like to navigate the 90-90 rule in an industry it dominated for the better part of a century.
 It's not new to the driverless game. It's been researching autonomous systems since at least 2007, when it competed in a government backed program exploring the technology. More recently, it bought San Francisco startup Cruise Automation to better understand artificial intelligence and aid the effort to integrate futuristic robotics with traditional car building.
 “This is quite frankly probably the hardest engineering problem in our time,” GM President Mark Reuss told me at the Journal's Future of Everything Festival this week. Most of the government funding for this research dried up long ago, and the war for talent and resources is fierce. To fund its drive, GM has killed several unprofitable vehicle lines and abandoned money-losing markets, resulting in better margins but lower revenues.
 In 2018, GM spent 12% of its $133 billion revenue on research, development and capital expenditures, an eye-popping figure for auto makers that shareholders would have thought imprudent just a couple of years ago. Wall Street has grown relatively receptive to these investments, but it's unclear how long that patience will last.
 I asked Mr. Reuss what last year amounted to a $16.5 billion question: When will GM's autonomous cars be ready for sale? He wouldn't put a date on it. “We're not going to sell autonomous vehicles in the near term” was as specific as he'd get. Like Waymo, GM sees its ability to marry the best sort of driverless operating system with the best riding experience at the lowest price as the Holy Grail. And, like Waymo, it is staging tightly controlled tests to make that happen.
 Larry Burns, a former GM executive who more recently has worked as an adviser for Google, lays out of the evolution of these strategies in his book “Autonomy.” Many analysts say GM is ahead of its conventional rivals in this type of research.
 Still, there is a significant difference between the GMs of the world and tech giants who would be content to see traditional auto makers serve as contract manufacturers. Waymo is owned by Alphabet Inc., a company with $109 billion in cash reserves, an advertising profit machine in Google and a seemingly bottomless appetite for money draining moonshots.
 This rivalry presents a bigger question that Mr. Reuss can't answer during a morning Q&A. Will the auto maker see its driverless project through? Or will it end up in the same closet where executives stashed the EV1 electric car, the Saturn car division, hydrogen fuel cells, or the AUTOnomy skateboard?
 The autonomous car holds great promise—with the hope of nearly eliminating the world's 1.25 million traffic fatalities each year putting it in the same league as advanced nuclear reactors, custom cancer vaccines or space travel as among the innovations under development that could most dramatically reshape our society. If GM plays a role in commercializing the technology, it would be a win for the company and for American manufacturing. In the Chevrolet Volt, recent history gives reason for skepticism, however.
 Hitting the market in 2010 as GM emerged from bankruptcy protection, the Volt was billed as a technological breakthrough. The plug-in hybrid vehicle was supposed to win GM widespread respect. Instead, it was squashed by Tesla Inc.'s electric sedans that went on sale shortly after the Volt's debut. Tesla sold 50% more cars in 2018 alone than the Volt sold over a nine-year production run, which quietly ended in February.
 An army of seasoned engineers, a century of experience and an $8 billion R&D budget wasn't enough for the Volt to beat Mr. Musk, a headstrong entrepreneur with a big vision and a mountain of venture capitalist cash.
 GM is rebooting its electric-vehicle strategy, Mr. Musk is scrambling to keep Tesla afloat, and the driverless-car puzzle remains unsolved. The scary part for GM is that this time Goliath is the other guy. Waymo's parent, Alphabet, has an $800 billion market valuation that is about 16 times bigger than GM's. The Mountain View, Calif., company has more than enough cash in the bank to buy it and Tesla at the same time.  

Lane Change
GM, once the No. 1 car company in the world, needs to shift focus from being the biggest seller to the biggest spender if autonomous cars are ever going to become mainstream.


Source: the company

BY JOHN D. STOLL

Add new comment