When the autonomous taxi company Zoox unveiled its vehicle last December it was somewhat of a surprise.It was a tall, squarish, perfectly symmetrical vehicle with rounded corners and sliding glass doors on either side and wheels with striking silver hubs. Inside was a large, open cabin with four comfortable-looking seats facing each other; the electric motor hidden beneath the floorboard could propel the vehicle in either direction at up to 75 mph.
The vehicle looked like a clean break from 135 years of automotive history. Which make sense: Not only is it self-driving, but it’s not not designed for sale to consumers. It’ll be part of a fleet. Zoox’s real product is a system of autonomous taxis that will move people around urban environments.
Zoox is one of the few companies that has built an autonomous vehicle from the ground up. Much of the work going on in autonomous vehicles today involves building autonomous driving software into existing consumer market electric vehicles. At seven-year-old Zoox, the plan was always to build a vehicle that could serve as an autonomous taxi in dense urban environments. After the company was acquired by Amazon in 2020 the vision remained the same.
But even though Zoox had a clear idea of its business model, it took a lot more thinking and design work to get to the eventual design that made sense for a taxi that could drive itself. The company is already testing its vehicles in San Francisco, and will be adding another test route in Los Angeles next year. In the future it will further evaluate them on test tracks, then private roads, then college campuses, the company told me earlier this year. The company hasn’t announced when its taxi service will launch.
Starting out—then starting over
Though Zoox was founded in 2014, it wasn’t until 2016 that the company really started to home in on the design it revealed last December, says Chris Stoffel who leads the industrial design group at Zoox. Before Stoffel arrived in April 2016, the company had been working from a basic design brief that defined what Zoox’s vehicle might be in broad terms. That left a lot of room for a wide variety of ideas of what the vehicle could look like.
“We have hundreds of sketches of different types of things and different types of environments, because there was no definition of what an autonomous vehicle for denser environments was” Stoffel says.
At the time, Zoox’s design team was smaller and the company hadn’t completely firmed up an agreed-upon set of design principles for the automobile. “We didn’t have a very strong engineering team, we didn’t have a strong experience team, we didn’t have a strong design team at the time and we were all growing,” says Stoffel. That fact led to an over-indexing on the wisdom of voices from outside the company.
“There was a lot of outside input at the time and it was a lot of people saying ‘no you can’t do that’ or people suggesting traditional ways, and I think that kind of led to this more traditional looking vehicle,” he says. But what the Zoox design team was after was a design that fully embraced and “celebrated” the autonomous part of the company’s vehicle.
Then Zoox started hiring people who had experience building products from the ground up—as opposed to iterating existing products—at other companies. The engineering, experience, and design teams began to grow. The new people were more apt to take fresh approaches to design than default to more conventional answers.
So in April 2016 Zoox threw out much of its design thinking from the recent past and returned to some of its first concepts of what the vehicle should look like. “We took a hard step back and said we need to change this—we need something more unique to us, something that celebrates the technology, so we reimagined it from the ground up,” Stoffel says.
One of Zoox’s first new design sketches used the concept of a “modern carriage.” It’s the idea of a compact horse-drawn carriage from the 1800s that has seats that face each other. Zoox’s vehicle, the sketch proposed, would borrow those themes, and use a powerful electric motor instead of a team of horses. The resulting design felt right, and seemed to respect the total new-ness of the vehicle Zoox was creating. For passengers, it created a social space that simply isn’t possible inside a vehicle where one of the seats had to be reserved for a driver who had to watch the road.
“When I got on board in April of 2016, that week we decided to change and move to that key sketch, to that design, which was really unique,” Stoffel says. “And from that day that’s been a very clear vision and a very clear path forward to bring that type of design to market.” “The sketch” is now part of Zoox company lore.
The location of the sensors on the vehicle also played a big part in the eventual design. Zoox’s engineering team was adamant that the sensors that serve as the eyes of the autonomous driving system be high up off the ground. The vehicle is a bit over 6 feet 4 inches tall—for comparison, a Toyota Prius is about 4 feet 10 inches—and the sensor pods extend from the top corners. This increases the vehicle’s overall view of the road, the engineers argued, and gives its autonomous driving system higher quality visual data to work with.
The tallness requirement had some serendipitous effects on the design. It increased the vertical space within the cabin of the vehicle. It also created room for the large doors on either side. Stoffel said that’s an important part of the design because Zoox wanted a vehicle that was easy to get into and out of.
It was also important to Zoox that the vehicle have a small footprint so that it could easily get in and out of tight urban spaces. (the bi-directional motor helps there, too.) The small footprint combined with the tall roof gives the vehicle a boxy look. But that look is the end product of a friendly tug of war between Zoox’s engineering and design group, between function and form, safety and style.
“There’s always healthy tension in any of those environments,” Stoffel says. Stoffel says this tension may occur between the industrial design group and the user experience, hardware engineering, and software engineering groups. The location and design of the sensor pods was a source of some of that productive debate.
But there were some good reasons for honoring the wishes of the software engineering group. It was important to put the software in pods that could be easily accessible in case a superior sensor technology emerged and Zoox wanted to build that onto its vehicles.
“That’s an evolving technology, absolutely, and they’re continuing to develop that as we move to bring a product to the public,” Stoffel says. “What that meant for us in design was ‘OK, we can’t constrain that; we have to enable flexibility for the future.'”
Stoffel says Zoox wanted the design process to be an open one, too. At many large companies, he points out, the various teams that work on different parts of a design often never get to see the whole product until the very end of the process. This could be for internal political reasons (stay in your own lane!), or it could be for maintaining the secrecy of a yet-to-be announced product.
“To me that always felt really weird, like I’m working on this car and I don’t get to be a part of it,” says Stoffel, who worked at Honda and Tesla before heading to Zoox to lead the design team.
So he decided to open the design studio to the entire company. “I said, look, we’re all working on this one product together–let’s make sure people can come in when they want,” he explains. The only rule was that there must be someone from the design team in the studio to explain what visitors were looking at and provide context.
The Zoox vehicle’s design is only one aspect of a project with myriad elements, some of them dauntingly difficult. The autonomous driving industry as a whole has slowly come to grips with the technical, regulatory, and cultural difficulties of self-driving cars. Some of the sparkle has worn off autonomous driving as a new, disruptive money-maker for tech companies. Developing the systems and the vehicles requires deep R&D pockets, lots of talent, and patience.
Amazon has said that it intends to leave Zoox to persue its pre-acquisition roadmap. But neither company has much else to say about their working relationship or possible future collaborations. Stoffel declined even to say whether Amazon’s executives had even visited his “open” studio.
Amazon has been exploring various types of autonomous delivery technology over the past few years.
IDC Next Generation Auto and Transportation research manager Matt Arcaro points out in an email message that one of the biggest hitches to that ambition is getting products over the “last mile” hop from a distribution center to the customer’s home. Right now half of Amazon’s delivery expense goes to that. And a big chunk of that last-mile expense is paying a driver, he says. Amazon has been exploring various types of autonomous delivery technology over the past few years, and Zoox may be a bet on another approach to solving that problem.
“If [Zoox’s] technology and expertise matures, this base foundation will allow a more seamless expansion (if the robo-taxi strategy is successful) or even a pivot (should the robo-taxi ambitions not meet expectations) to support new domains like package delivery and logistics,” Arcara says.
Some observers said that the $1.2 billion selling price on Zoox was a bargain for Amazon. But Zoox, which ran short of funds during the pandemic, needs time more than it needs money, and that’s what it got.
“The race to full autonomy is a marathon,” Arcaro concludes, “And those looking to win need to commit for the long haul.”