After more than a century making vehicles for humans to drive, General Motors has ripped the heart out of its latest ride, and is now holding the grisly spectacle up for all the world to see: A car with no steering wheel. And it plans to put a fleet of these newfangled things to work in a taxi-like service, somewhere in the US, next year.

And no, this robo-chariot, a modified all-electric Chevrolet Bolt, doesn't have pedals either. This is GM's truly driverless debut, a car that will have to handle the world on its own. No matter what happens, you, dear human passenger, cannot help it now.

Terrifying? Maybe. But it's also a major step in GM’s aggressive bid to maintain its big dog status as the auto industry evolves away from individual ownership and flesh-and-blood drivers. And it’s just the beginning for the Detroit stalwart. “We’ve put together four generations of autonomous vehicles over the course of 18 months,” says Dan Ammann, GM’s president. “You can safely assume that the fourth generation won’t be the last.”

While Waymo, Uber, and others in this space are building webs of partnerships to deploy their autonomous tech, GM is going for vertical integration. It spent a reported $600 million on Cruise Automation, and put the San Francisco–based startup in charge of its effort to develop fully autonomous vehicles. It bought its own lidar manufacturer: Strobe, a Pasadena-based outfit GM thinks could cut sensor costs (a big problem, especially for lidar) by 99 percent.

To lay the foundation for a business that doesn’t rely on selling cars to people, it launched a car-sharing service called Maven. And it has flexed its manufacturing muscles (perhaps its biggest advantage coming in), configuring its Orion assembly plant north of Detroit to build this latest generation of robocar. Indeed, GM is counting on its manufacturing prowess to give it an edge in this new world. “Either you can do that or you don’t have a business,” says Kyle Vogt, who founded and heads Cruise.

And yes, now is the time for this autonomy stuff to really become a business. The technology has made massive leaps in recent years, and not just chez Cruise: Waymo, which started as Google's self-driving car, plans to deploy similarly driverless cars in the next few months. These systems aren't perfect, and they won't work everywhere or all the time. But the companies behind them are ready to press ahead.

Of course, dropping the steering wheel gets tricky when you consider federal regulations that require things like steering wheels. So GM has officially asked the federal Department of Transportation to exempt these vehicles from certain parts of the rules that govern automotive safety. Because those were written for human-driven cars, they include requirements like a foot-activated brake pedal and an airbag built into the steering wheel.

In an age where cars won’t need any kind of pedals or steering wheels, those don’t quite make sense. They’re “almost illogical or missing a predicate when there is an artificial intelligence, a computer driver,” says Paul Hemmersbaugh, GM’s policy director for autonomous vehicles. Pending legislation would let the feds grant up to 100,000 such exemptions a year for each manufacturer, up from the current 2,500. Good thing, because there's no serious movement to rewrite the rulebook.

The federal government in general is all for autonomous vehicles, and usually grants such exemptions, so that’s one of the easy bits. Harder is finding the right spot to launch this system. Vogt wouldn’t offer any clues (nor would he say how many cars will make up the fleet), but you can put your money on one of the states that have expressly welcomed self-driving tech without much in the way of rules. California, where companies must publicly report crashes and other data, seems less inviting here than a state like Arizona, which doesn’t put any special restrictions on robocars. Plus, the weather in Phoenix is good year round (minus the occasional haboob), and the driving environment is far simpler than a place like San Francisco, where Cruise does the bulk of its testing.

The real tough part, however, will be accounting for the wildcard: the human passenger. In the past century, GM’s relationships with its customers more or less ended when the dealer handed over the keys. It never had to think much about how people behaved inside its vehicles. Now it does. Vogt says his team has considered how to account for all sorts of annoying human habits. If the rider doesn’t close the door after walking away, the car can do that itself. But plenty of questions—like what the car should do if it can’t safely and legally pull over near its passenger’s pickup or drop off point—remain.

To handle riders who demand a human touch, and to do things like call emergency services in case of a crash, GM will rely on its in-vehicle OnStar system. And, as an early test for a rideshare system, GM built an app with which Cruise employees can call robocars for free rides around San Francisco. It’s a logical start, but challenges will emerge that engineers and human factors specialists would never think to consider.

The old-school behemoth had better remain flexible enough to navigate them—steering wheel or no.


The Road to the Future

Read more: https://www.wired.com/story/gm-cruise-self-driving-car-launch-2019/