The prototype Zoox vehicle is seen doing a photo shoot in San Francisco
Zoox, the self-driving robotaxi startup acquired for over $1B earlier this year by Amazon
Zoox’s mission, from day one, was to design a custom vehicle from the ground up to be a robotaxi and only a robotaxi, forgetting all conventional wisdom about cars. Teams like Waymo have instead modified existing vehicles for their first deployments. GM/Cruise also began with modified Bolts but later displayed a vehicle with superficial similarity to the Zoox, and Zoox of course has been doing all their on-road testing in modified standard vehicles while developing this vehicle in parallel.
In the past, Zoox has hinted at design goals for this vehicle. As predicted it is symmetrical, and has no front or rear. It is intended to be able to go in both directions equally well. Some hints from the company say the 4 wheels can also be steered independently, allowing it to be highly maneuverable, possibly able to turn in place as well. The original design called for total 4-way symmetry, with one motor for each wheel and complete redundancy. Sources suggest this has been scaled back a bit.
This was part of a giant week of Robocar news, including:
- AutoX no-driver service in Shenzen
- Waymo opening up taxi service to the general public near Phoenix
- Cruise does first test rides with no safety driver in San Francisco
- Aurora merging with/acquiring Uber’s
UBER ATG self-driving unit - Some no-driver tests by Baidu
BIDU BIDU
This basic robotaxi shuttle design, from the outside, is very similar to others from Cruise, Easymile and many others, with sliding doors in the center of a rounded boxy shape, and face to face seating. There’s only so much variation possible there. Notable about this vehicle are the 8 LIDARs located at the corners, one horizontal and one tilted down. Unlike most other vehicles they are on mounts sticking out of the car, giving them a slightly better view and the ability to even see the sides of the vehicle itself with no blind spots, useful in close quarter work and scanning nearby pedestrians and passengers. (The Cruise Origin also has the LIDARS extending out from the vehicle.)
An early reveal of Zoox's mule prototype shows 4 wheel steering
Last year, I asked Jesse Levinson, co-founder and CTO of Zoox about some of the reasons he wanted to take on the hard task of making a custom vehicle. While there are several factors, one he outlined was indeed the ability to have complete control of sensor placement. Tim Kentley-Klay, the ousted other co-founder, particularly valued the symmetric design for ease of manufacturing and for rider convenience. Such a vehicle never “backs up” because it has no forward or backward direction. As such, it can pick up customers and immediately take them away without needing to turn around. It remains to be seen if that’s a killer feature or a nice-to-have. (Vehicles can also usually do their backing up before they pick you up, since robots don’t really care much about the hassles of turning around.)
Zoox's earliest designs show a symmetrical vehicle with no windshied. Fortunately they gave that ... [+]
Zoox’s commitment to this design may have cost it dearly. Such an ambitious plan was costly, and for various undisclosed reasons, Zoox ran out of money and needed more earlier this year, but was unable to raise it during the Robocar winter. Instead it sold itself to Amazon at a bargain price. Most people imagined Amazon wanted the company for its team, to build technology for Amazon’s massive logistics business, but Amazon has said it wants to continue the robotaxi vision (though of course that doesn’t mean it isn’t also interested in logistics.) The upcoming reveal indicates some devotion to that plan. It’s not so surprising. As big as Amazon is, the robotaxi business is even bigger if you can win a big part of it.
The design is not tall enough to stand up in, but at the same time is not tremendously aerodynamic. Most electric vehicles that will travel at high speed work very hard to have an aerodynamic form — it’s the only way they can get good range. This suggests this vehicle is an urban one, not meant to do long trips on the highway, though Zoox says it can do 75mph. It also has independent 4 wheel suspension, which should allow it to have a very smooth ride, particularly when combined with LIDAR scans of the road surface ahead. Another design (like the very early one above) might be more suitable for long trips. (The nice thing about robotaxis is you can tailor the vehicle to the trip.) Not going on the highway at all can eliminate the need for long-range LIDAR, though there are two boxes on the top of the vehicle which might be such a LIDAR — the picture does not have enough detail. The vehicle will almost certainly also have radars (front and back, of course.)
Let’s look forward to Zoox providing us more details on Dec 14.
