Uber is bringing self-driving cars to Pittsburgh this month. Sort of.

For the past several years, Google and other companies have been testing out self-driving car technology on public roads. But if you were a member of the general public, there was no way for you to ride around in a self-driving car.

Uber is about to change that. Later this month in Pittsburgh, the ride-hailing company will offer the first self-driving car service that’s available to the general public. It represents a big step toward transforming self-driving cars from a research prototype into a commercial service.

Yet the service also comes with a big asterisk. While the new program will be open to the general public, it’s still very much a work in progress. Rides will initially be free, and they’ll still have a driver behind the wheel making sure nothing goes wrong.

This means Uber CEO Travis Kalanick is still far from his dream of offering a fully autonomous ride-hailing service that could slash prices and boost Uber’s profits. Rather, the new service reflects the start of a new project to make sure Uber doesn’t fall behind its rivals — especially Google — in self-driving technology. It will be years before you’ll be able to hail an Uber and have it show up with no driver inside.

Uber has invested heavily in self-driving technology in the past 18 months

Photo by Justin Sullivan/Getty Images

Kalanick has long seen self-driving cars as the long-term future of his company. In early 2015, he began to make it a priority. And he decided to base Uber’s new self-driving research center in Pittsburgh so he could raid Carnegie Mellon’s top-tier computer science school for self-driving car talent.

In a matter of months, Uber hired about 50 researchers and scientists from CMU. The team began to apply what they had learned in the lab to the real world, building self-driving car prototypes. By May 2016, Uber’s self-driving car prototypes could be seen driving around on Pittsburgh streets.

Extremely accurate maps are essential for a fully self-driving car, so in 2015 Uber acquired a mapping startup to help build its own maps and began building detailed maps of Pittsburgh’s urban landscape. Uber plans to dramatically scale up that project, spending $500 million to build maps around the world.

A big challenge for developing self-driving car technology is to collect massive amounts of real-world data. Collecting millions of hours of real-world data provides the raw material that software engineers use to refine their self-driving car software, finding and weeding out bugs that could lead to a fatal crash.

Google has had to do this the hard way, hiring dozens of test drivers to ride around the San Francisco Bay Area (and more recently Austin and other cities) while watching carefully to make sure the car doesn’t crash.

Uber’s existing ride-hailing business should make this process more cost-effective for the company, since it will be able to ferry passengers around — earning some revenue — while collecting test data.

But Uber evidently doesn’t think its technology is quite mature enough for that yet. According to Bloomberg, Uber’s “self-driving” cars will have not only a safety driver ready to grab the wheel at a moment’s notice but also a “co-pilot” in the front passenger seat monitoring the car’s software to make sure nothing goes wrong. Right now, for example, the human driver has to take over when a car is crossing a bridge. In other words, this is still very much a research project rather than a service that has any hope of making money.

Presumably, Uber’s plan is to gradually change that, replacing the Uber employees who will serve as the initial test drivers with ordinary Uber drivers, and then eventually eliminate the drivers altogether. Uber’s vast database of both customers and drivers means the company will be able to scale up the program quickly once it’s convinced that it’s safe to put ordinary Uber drivers behind the wheel.

Of course, the barriers here might not only be technological. A variety of state and federal regulations could prevent Uber from entirely dispensing with human drivers. Uber — along with other companies working on self-driving technology — will likely have to convince policymakers to modify or reinterpret these regulations before it will be allowed to send anyone a car with no driver at all.

The race to build a fully self-driving car keeps getting more intense

Uber isn’t planning to build cars itself; its plan is to collaborate with conventional carmakers, which will make the actual vehicles. On Thursday it announced a new partnership with Volvo. Uber’s first self-driving cars will be Volvos modified with Uber’s technology, and the two companies plan to spend $300 million to jointly develop a new self-driving car model. The two companies are aiming to have the car on the road by 2021.

That date is significant because Ford also recently vowed to bring fully self-driving cars — cars without steering wheels or pedals — to market in 2021. That’s roughly consistent with statements executives from other companies — including Tesla, Google, and major carmakers — have made about their own plans for self-driving cars.

Of course, overconfidence is common when bringing ambitious new technologies to market, and few efforts are more ambitious than fully autonomous cars. It’s likely that some of these projects will fall behind schedule, and others may wind up being only partly autonomous, with drivers having to stay ready to grab the wheel in case of emergency. (This is how Tesla’s Autopilot feature works now.)

Still, with billions of dollars being poured into self-driving car research, companies have every reason to worry about being left behind. So lots of companies are now making ambitious goals to bring self-driving technology to market by the early 2020s.

And Uber’s position in the market gives it a big advantage: Because it doesn’t make cars itself, it can partner with multiple car companies, play them off each other, and ultimately work with whoever gets self-driving technology to the market first. Uber’s deal with Volvo is nonexclusive, and the company may sign similar pacts with other carmakers in the coming months.

No matter who makes the best self-driving cars of the 2020s, there will be millions of consumers who have grown accustomed to hailing rides using their Uber app. Hence, it will likely be an easy sell for Uber to convince those customers to use the same app to hail a self-driving car.

Will you support Vox’s explanatory journalism?

Most news outlets make their money through advertising or subscriptions. But when it comes to what we’re trying to do at Vox, there are a couple of big issues with relying on ads and subscriptions to keep the lights on:

First, advertising dollars go up and down with the economy. We often only know a few months out what our advertising revenue will be, which makes it hard to plan ahead.

Second, we’re not in the subscriptions business. Vox is here to help everyone understand the complex issues shaping the world — not just the people who can afford to pay for a subscription. We believe that’s an important part of building a more equal society. And we can’t do that if we have a paywall.

So even though advertising is still our biggest source of revenue, we also seek grants and reader support. (And no matter how our work is funded, we have strict guidelines on editorial independence.)

If you also believe that everyone deserves access to trusted high-quality information, will you make a gift to Vox today? Any amount helps.

One-Time Monthly Annual

$5/month

$10/month

$25/month

$50/month

Other

$

Yes, I'll give $5/month

Yes, I'll give $5/month

We accept credit card, Apple Pay, and Google Pay. You can also contribute via