There’s nothing normal about riding in a vehicle that actually, really drives itself.

The sensation only intensifies when there’s no distinguishable front or back end, no steering wheel, brake, or accelerator pedals, and the exterior more closely resembles a microwave than a car. The Zoox robotaxi ticks all these boxes and more.

I got a chance to ride with the Amazon subsidiary's CTO and co-founder, Jesse Levinson, around Las Vegas during CES. It was an eye-opening look at one of the more interesting and ambitious efforts in the self-driving space, and one that could be a competitor to Google's Waymo—if a lot of things go right first.
Rolling Out

The boxy vehicle rolls up to the curb, and a technician opens sliding side doors (not unlike those on a bus or New York City subway) using an iPad. Levinson, me, and a PR person pile into the spacious little vehicle, each of us sitting in one of four firm seats facing the middle of the vehicle. It’s a remarkably communal space, with plenty of room for passengers to cross their legs comfortably without awkwardly bumping into someone (that is, as long as passengers don’t have suitcases with them. The Zoox Robotaxi has no trunk space.)

The seats are slightly cushioned, an upgrade from the very firm first-generation Zoox vehicles, a few of which sit in a Las Vegas warehouse slowly being used for parts. Because the seats sit at either end of the square vehicle, there’s very little front or rear visibility out of windows high above passengers' heads at the front and rear of the vehicle. You get more views out of the mostly glass side doors, but it’s difficult to see what kind of potential road hazards lie ahead given the layout.

Levinson launches into an explanation about why he got into robotaxis more than 10 years ago, noting that individual vehicle ownership is “really inefficient.” As Levinson said, “People don't really think about that because we're so used to just car culture,” calling out the fact that there are more than two cars per family in the U.S., and 96% of the time they sit around taking up space in someone's garage, rather than being used.

“It's really wasteful, not just economically but also environmentally," he said. "The question is, can we use our resources more efficiently? The reason why Zoox is compelling to me is that we have the opportunity to both make a way better customer experience for moving around our cities and really utilize our resources on this planet more efficiently, all while saving lots of lives from a safety point of view."
The Ride: Safe And Overly Cautious Like A New Driver

Once you’re boarded, you start the ride by using any touchpad located in the interior of the vehicle, near each of the doors. You can control everything from the audio and lighting to controls for pulling the vehicle over or checking how much longer your ride might take. But that’s where direct human control stops. The rest is up to the little toaster ushering you along at the speed limit, jouncing over potholes and road grime while making some very conservative—and somewhat beginner-level—driving decisions.

The first such behavior happened in a construction zone. We were riding along in the right-hand lane when the vehicle stopped rather abruptly. It had driven right up to where the lane was physically closed and activated its left signal, waiting patiently for a gap in traffic. Since the Zoox isn’t exactly efficiently shaped, it needed a relatively large gap before changing lanes, which it did eventually, and proceeded to the next traffic light.

Had a human driver been navigating the construction zone, they would have (hopefully) looked well enough ahead to make a move far sooner than what actually happened. That would’ve prevented any awkward merging, which could be very problematic during rush hour traffic.   

The Zoox has a 133 kW battery pack, and multiple redundancies so there is  “no single point of failure” according to the company, meaning that if, for example, one of the motors or drive units fails, the others can take over.  The vehicles can run for about 16 hours before needing to charge and can safely reach 75 mph, according to the company, though wind noise and comfort at those speeds might be a bit to reckon with. We likely didn’t go more than 50 mph since that was the speed limit on the roads we traveled, but that’s just an estimate. There is no accurate way to know how fast you’re going in a vehicle with no steering wheel or gauges.

“We've been able to put more safety and more redundancy and more customer features in this vehicle, knowing that we don't have to be able to sell it for a consumer price point,” said Levinson. “We think that's a really great product and a great service, and the best way to get this technology out there safely.”

During the pre-planned route (which the vehicle had already driven numerous times), we stopped at a traffic light where a pedestrian crossed the road. We were first at the line; the Zoox waited patiently for the pedestrians and the light to change and we proceeded straight ahead, no harm or foul. However, I do wonder what would have happened had we needed to make a right turn.

Would the Zoox have waited until the pedestrian got out of the way before making a legal right on red? Or would it have acted like that same newbie driver in the construction zone, waiting for a green light while annoying everyone stacked up behind?

As it was, we carried on our way with a few more slightly aggressive stops after a traffic slowdown, like a New York driver who doesn’t know how to modulate the pedals and is surprised by a suddenly stopped car. Our path took us from the Zoox warehouse on the edge of Las Vegas to the back part of the Las Vegas Strip and took roughly 30 minutes to complete.

Arguably, the true test of an autonomous vehicle without controls is how many times it makes passengers wince, cringe, or tense in a single session. The Zoox didn’t once make me nervous, though the abrupt stops, rough ride over potholes, and the rookie-style decision making made me slightly uncomfortable. It’s more like a tentative, somewhat frightened teenager learning how to navigate the real world, aggressively applying the brakes on occasion at stop signs or traffic lights. Sure it jostled us around a bit, but it was far more comfortable than the ride I had in a Lyft on the way to the airport, in a vehicle with broken shocks. In this case, there was no driver who insisted on turning around to talk to me, and did not pay attention to the road ahead.

The ride inside the Zoox is not what you’d call cushy, by any means. With the wheels pushed to the corners and the ride height so low, uneven surfaces translate readily into the cabin. The seats, though they are upgraded from the previous version to be softer and more comfortable, are still very firm. The interior offers plenty of space for passengers, but if you packed four people in there with luggage (there’s no trunk, obviously), it would get tight. Plus, if the luggage included roller bags, they’d be everywhere with the way the vehicle lunged at stops.
Safety First?

Levinson says that the reason for the lungy stops and newbie behavior is to prevent crashes and collisions.

“The last few years at Zoox, we have been really focused on the safety case,” he explained. “You know, how do we put these vehicles on public roads with no safety driver and make sure they're safer than a human? Now, it's about expanding those geo fences, adding lots of great pick up and drop off locations, getting our first customers, but also making sure we can do that in a sustainable and economical way.”

Zoox has been operating test rides and routes in the purpose-built toaster-shaped vehicles in Las Vegas, Foster City and San Francisco for the last few years. The company says it will open up semi-public rides in the microwave-style vehicles sometime in 2025, but those rides will only be available to a very limited group of employees, their families and friends, as well as a group of “Explorers” who gain access on an invite-only basis.

There's no concrete timeline on when the service will be available to the wider public, and Zoox wouldn’t share how they plan to choose these Explorers, for the test rides in the tiny handful of cities and locations where you can actually hail one.
A Bumpy Road Ahead?

In addition to the toaster-set of taxis, Zoox also has a fleet of retrofitted Toyota Highlanders that serve as test beds for the future bespoke vehicles. These don’t carry passengers and are for testing only, but that testing has not been without safety incidents. The company has been under investigation by the National Highway Traffic Safety Administration for collisions that its specialized Highlanders had with motorcycles near its headquarters in San Francisco. Just last week, it had to recall 258 of those Highlanders to fix a sudden braking issue.

Zoox is also working through some issues that have arisen with its self-certification, which the company argues all automakers do. Zoox, however, is the only self-driving company that does so, which has made regulators nervous.

There’s also the question of whether or not it makes sense to build these vehicles as bespoke options, especially against the backdrop of the current autonomous taxi environment and the general mayhem that the current administration seems set on causing in the automotive sector.

The autonomous cab space is expensive, very small, and shrinking. Just last year, GM’s Cruise closed up shop. Argo.AI closed in 2022. Elon Musk says his Cybercab will launch in Austin this year, but as the world knows, the gap between his promises and reality is a wide one.

Alphabet’s Waymo is currently the only autonomous taxi company that is making real headway, and the company has a partnership with China’s Zeekr for their own bespoke people-mover vehicle. But that has been delayed significantly (it was due to launch in the US in 2024.)

While the Zoox experience still feels far from a human driver experience, the company has made slow and steady progress toward its autonomous taxi goal. Whether or not it will be the future that Levinson and Amazon hope it will be is another story altogether. That’s especially true given the backdrop of current political and economic upheaval. Yet Levinson is upbeat.

“I think it's great that there are multiple approaches out there,” he said. “Our personal view at Zoox is that the types of engineering and design and business choices you make—with a car that's designed to sell to customers, versus a robotaxi that is really designed to be owned and operated as a fleet—are pretty different. We'll see what happens with our fellow travelers.”