Waymo One, the groundbreaking self-driving taxi service, explained

Today is a day that fans of self-driving cars have been anticipating for years. Waymo—widely seen as the industry leader—is finally launching its “Waymo One” commercial taxi service in the Phoenix metropolitan area.

The announcement fulfills Waymo’s long-standing promise to offer a commercial service by the end of the year. But the launch comes with important caveats.

Initially, the new service will only be offered to Waymo’s early riders—the same handpicked test passengers that have been riding in Waymo’s vehicles for the last 18 months. Waymo says it hopes to make the service available to the broader public “over time.”

And despite the fact that Waymo began testing fully driverless vehicles more than a year ago, Waymo has chosen to put safety drivers behind the wheel of its cars for the commercial rollout. That suggests Waymo still has doubts about whether its technology is ready for fully driverless operation.

Of course, there’s nothing magical about Waymo’s self-imposed 2018 deadline. What matters is what happens in the coming months. Can Waymo phase out the safety driver, scale up its service, and make self-driving taxis broadly available to the general public? Or will technical challenges hobble Waymo One’s growth, making it little more than a rebranding of Waymo’s existing rider testing program?

Under the early rider program, Waymo required participants to sign non-disclosure agreements. Those restrictions will be lifted for customers who use Waymo One, allowing them to take pictures and videos and to share the information on social media. In the coming weeks we’re going to learn a lot about how Waymo’s technology works—and doesn’t work—in the real world.

Waymo is competing with Uber and Lyft

Waymo’s self-driving technology is innovative, but its business model is completely conventional. Waymo One will be a ride hailing service a lot like Uber or Lyft.

Passengers will hail vehicles with a Waymo One app, which will be available for both iOS and Android phones. Like the Uber and Lyft apps, it prompts the user to enter a pickup location and a destination. The app provides estimates of the cost and likely arrival time before the customer books the ride.

Credit: Waymo

Fares are based on time and distance, and customers can expect fares to be roughly on par with what you’d pay for an Uber or Lyft trip—perhaps even a bit lower. The above Waymo-provided screenshot shows a customer booking this trip, which is 4.6 miles long and takes about 12 minutes. Waymo charges $7.32 for the trip. I punched the same route into Lyft and Uber aps on Tuesday afternoon and got quotes of $8.29 and $9.38, respectively.

Service will be available in portions of four municipalities southeast of Phoenix: Chandler, Tempe, Mesa, and Gilbert. Arizona State University is based in Tempe, while Intel has a major presence in Chandler. These are sprawling municipalities with a combined area of more than 300 square miles (and a combined population of over a million people), but Waymo will initially provide service in less than 100 square miles of that area.

Waymo has pre-selected recommended pickup and drop-off locations throughout the service territory—they’re marked with a “W” icon in the screenshot above. Waymo vehicles prefer to drop customers off in these locations, which means that customers may need to take a short walk at the start and end of a trip. But Waymo says that—as with Uber or Lyft—customers will have the option to override these suggestions and request pickup or dropoff in another location.

While the service officially launches today, most residents of the region won’t be able to download the app yet. Waymo still has a fairly small fleet of vehicles, and there’s likely to be a surge of interest in this new technology. So to avoid getting swamped (and suffering negative coverage for long wait times), Waymo as noted is limiting initial sign ups to people who are already in Waymo’s early rider program. In the future, Waymo plans to let more people into the program, with the goal of eventually offering access to everyone in the service territory.

What it will be like to ride in a Waymo car

Customers will hail a car on their Waymo One app. Once it arrives they’ll climb into the backseat. For at least the first few weeks—and possibly longer—there will be a safety driver in the driver’s seat who will monitor the car’s performance and answer customer questions. Customers also have the option to call customer support by pushing the “help” button on the overhead control panel, located on the ceiling above the middle-row seats, or they can request a call from a Waymo service agent from within the Waymo One app. These options will become more important when Waymo eventually starts operating cars without safety drivers.

Credit: Waymo

Once customers are ready to go, they push the blue “start ride” button on the same overhead control panel.

A pair of touchscreens on the backs of the front-row seats keep passengers updated on the ride’s progress. Passengers choose between two different views. One view shows a map of the route and an icon to mark the car’s current location.

Credit: Waymo

The other view shows a schematic view of the car’s surroundings, taken from a perspective above and behind the car. Pedestrians, cars, buildings, and other significant objects are shown as simple geometric shapes. Every few seconds, a ghostly pulse briefly shows a more detailed view, with the outline of trees, lamp posts, and other nearby objects becoming visible.

The point of all this is to boost passenger confidence that the car is fully aware of its surroundings. Passengers will be able to look out the window and verify that the objects they see outside match the ones they see on the screen. Sometimes the screen will also help inform customers about what the car is doing—for example, if the car stops to let a pedestrian cross, the screen will make that clear.

The big question: Can Waymo maintain its lead?

Waymo isn’t quite the first company to launch a commercial self-driving car service—several startups did that earlier in the year. But today’s launch makes Waymo One the most technologically impressive self-driving service on the market by a wide margin. The few startups that have launched services are operating in much smaller, more tightly controlled environments and at lower speeds. Meanwhile, Waymo’s larger rivals—including GM’s Cruise, Ford’s Argo, and well-funded startups like Zoox and Aurora—are one to three years away from launching commercial services of their own.

Waymo has such a big lead because Google began investing in the technology almost a decade ago—long before most other companies were taking it seriously. The question now is whether Waymo can effectively capitalize on its early lead.

That would mean expanding rapidly to other cities, giving Waymo time to establish itself before rivals enter the market. In June, Waymo ordered 62,000 Pacifica minivans—a sign that the company is thinking ahead to markets beyond Phoenix.

But Waymo won’t be able to fully focus on expanding into new cities until it has mastered driverless operations in Phoenix. Hiring, training, and supervising safety drivers is expensive and time-consuming, so Waymo’s expansion will be seriously slowed—and likely rendered unprofitable—if it has to do that for every city where it offers service.

The big question, then, is why Waymo is launching its service with safety drivers after more than a year of testing fully driverless vehicles. It’s possible that the technology is very close to being ready and that Waymo is just using safety drivers out of an abundance of caution. After all, Waymo already has them on staff, so there’s little downside to using them at launch time.

But it’s also possible that Waymo’s technology is far from ready for fully driverless operation. An August scoop from The Information’s Amir Efrati provides the pessimistic viewpoint here: