Thanks for reading Hyperdrive, Bloomberg’s newsletter on the future of the auto world. A Drama-Free Driverless Ride |
The Zoox robotaxi has no steering wheel, no pedals and four inward-facing seats. Last week, I was the first journalist to hitch a ride in one of the vehicles in their eventual launch market: Las Vegas. It’s an opportune time to be getting one’s arms around the state of robotaxis. Elon Musk is staking Tesla’s future on the concept and has set a date to unveil a vehicle in August. But after years of hype about fully autonomous vehicles transforming transportation, it remains to be seen whether the technology is ready, and how quickly companies like Zoox will be able to scale it with thousands of cars on the road. Zoox’s car whisked me through a 5-mile loop of public roads not far from the strip across mixed, multi-lane traffic and at speeds of up to 45 miles per hour. The journey went much like any other ride-hail experience: Pull up the app on your smartphone, set the destination and confirm. Hop in, buckle up and press “Start” on one of the small control panels by each of the passenger seats. There’s air conditioning — a must in Sin City’s 100F (38C) desert heat — and you can play music. Beyond that, the interior is pretty bare bones. I was escorted on the ride by Zoox's co-founder and chief technology officer Jesse Levinson, who wanted to demonstrate to me that Zoox has been making quiet progress since Amazon acquired the company four years ago. Its long-held objective has been to operate an autonomous ride-hailing service — think Uber or Lyft, without human drivers — using purpose-built robotaxi the company plans to own and operate itself: no retrofitted consumer cars, and no partnerships.  Zoox’s electric robotaxi. Source: Zoox Zoox’s vehicle was familiar to me. I visited the company’s California headquarters in December 2020 and was the first to get a look at the car’s design and specifications. But this recent trip was my first time going for a ride in a non-controlled environment (Zoox already operates a limited 1-mile route between two of its facilities in Foster City, California.) The Las Vegas deployment will render Zoox the first company in the US to operate a purpose-built robotaxi on public roads and at relatively high speeds. Over the years I’ve covered Zoox and known Levinson and its CEO, Aicha Evans, they’ve had this very clear end goal. But unlike Alphabet’s Waymo or General Motors’ Cruise, Zoox resisted the urge to shout about commercializing its technology or offer set timelines for a commercial service. In another big contrast, Zoox decided against testing the business model or giving the public a taste of the service in retrofitted cars that do have both a steering wheel and pedals. During my journey, Zoox’s vehicle was able to conduct several unprotected right turns, where it had to yield the right-of-way to oncoming traffic and pedestrians before turning. Zoox’s car managed to discern that it was safe to turn right at a red light when there was no oncoming traffic. However, the vehicle didn’t attempt any left turns — protected or otherwise — during the journey. When I asked Levison about this, he said it was just the layout of the route and insisted that the vehicle is able to handle them. In the course of our trip, we experienced cars cutting in front of us, as well as construction along the median and lane closures. Out an abundance of caution, Zoox wouldn’t let us set out on our journey until a busted light signal was fixed. I’d liken the experience to being in an airport shuttle, or even on a monorail, with the biggest difference being that the Zoox can traverse several lanes of traffic and will change lanes based on the cars around it. I’ve taken a lot of Waymo autonomous ride-hail trips, and I’ve also tested the driver-assistance system that Tesla markets as Full Self-Driving in my Model Y. My experience of this one, 5-mile ride was that Zoox’s vehicle drives more smoothly. It makes quicker decisions and behaves much more like a typical human driver. One big psychological difference to get over is that you can’t see out of the front or back windows, by design. Nor are there any screens showing a digital visualization of what the cars sees — something that both Waymo and Tesla do offer. Zoox has an elaborate sensor suite on each corner of its car, comprising lidar, radar and cameras. The company designed a cheaper way to build its robotaxi and a simple assembly line in California. Most of the components, including drive units, arrive preassembled by suppliers. That’s an economic trade-off for Zoox, because while the per-unit cost of each robotaxi is high, its suppliers will be capable of building thousands of units per year. If those robotaxis can run for significant portions of the day and charge fares similar to what current ride-sharing apps do, then the company will quickly be able to cover its costs, and the business model will prove successful. Levinson argues that a purpose-built robotaxi, operated by its maker, is the safest way for the technology to be deployed in the real world. He believes it would take any company a minimum of five years to pull this off. “It takes a long time to design and build one of these things, and then to make sure that it’s safe,” Levinson told me during the ride. “Not only does nobody else have one of these out on the roads, we don’t know of anybody that’s even working on it right now. Of course, somebody could be, in secret.” In many ways, Waymo and Cruise seemed to pull ahead of Zoox the last few years. The former two deployed fleets of test cars and converted them into vehicles for revenue-generating ride-hailing services. Cruise’s mishandling of an incident in which one of its cars struck and dragged a pedestrian derailed its business, and safety regulators have subjected all three of the players to closer scrutiny. The triumvirate could get some company soon in the form of Tesla, which now appears to share Zoox's view that pursuing purpose-built robotaxis is the way to go. Musk is slated to unveil the company’s car on Aug. 8th. But there are significant differences between Tesla’s approach to developing self-driving capability from the others. Musk, for instance, has espoused a camera-only system for detecting surroundings, whereas Zoox believes autonomous driving is only achievable with a multi-sensor suite of lidar, radar and other inputs for a richer and more robust 360-degree view around the car. Zoox will soon be able to test its robotaxi in San Francisco, alongside the large fleet of retrofitted test cars it already has in the city. Other test-fleet locations are planned, including Austin and Miami, but the company will continue to be deliberately slow and steady in how it expands. “I’ve really stayed away from the race and, you know, calling who has pole position and who doesn’t,” Evans told me at the company’s Vegas facility. “What I will tell you is that Zoox has even more conviction in the path that it’s chosen." — By Ed Ludlow | 
