The day may soon be coming when robotaxis finally become a reality that the average American can experience.
General Motors–backed autonomous driving startup Cruise said on Tuesday that it has opened up its self-driving car fleet in San Francisco to the public.
While capacity is limited with only a handful of cars available, the opportunity to sit in the back seat of a car that steers itself as if by ghostly hand is no longer reserved for a select few engineers, test drivers, and GM chief executive Mary Barra.
“We’re opening a sign-up page on our site today so you can get a driverless ride soon—and free, for now,“ Cruise interim CEO and cofounder Kyle Vogt said in a post. “We’re starting with a small number of users and will ramp up as we make more cars available.”
In a letter to shareholders, CEO Barra said, “This major milestone brings Cruise even closer to offering its first paid rides and generating $50 billion in annual revenue by the end of the decade.”
This follows an announcement in July from Ford and Lyft that they will debut robotaxis in Miami using the know-how of self-driving startup Argo AI. More recently, Intel subsidiary Mobileye in September revealed its plan to start a similar commercial fleet in Germany this year. Meanwhile, Alphabet’s self-driving unit Waymo has already been chauffeuring passengers in Phoenix as part of its Waymo One service since 2020.
The race to develop autonomous driving technology has been a longer slog than many initially anticipated some five years ago. Tesla CEO Elon Musk famously predicted he’d already have 1 million robotaxis on the road by the end of 2020.
Yet engineering a car to drive on its own and receiving regulatory approval to introduce the technology on public roads, while crucial milestones, are only two variables in the equation.
A far more important question is whether a profitable business can be built around them since the economics are challenging even after the cost of a human driver is removed.
How to scale the costly technology is the final challenge.
Most robotaxis driving around are easily distinguishable by their characteristic roof equipped with all sorts of sensor arrays. Not only do these vehicles require the most powerful processing chips available, they also typically feature multiple lidar sensors at various corners of the car, including on top for a greater view.
Similar to radar, these devices emit pulses that bounce off objects to continuously map an extremely precise three-dimensional map of the vehicle’s surroundings.
They are the gold standard when it comes to safely operating a robotaxi, but they remain prohibitively expensive and a hurdle when building a fleet. Waymo One is still in the process of testing its cars in San Francisco, only its second city after Phoenix.
Ordinary individuals would balk at the price of a fully autonomous vehicle, so for now likely customers would be mobility service providers such as Uber and Lyft that operate a fleet.
At the same time it launched its commercial service, Vogt said Cruise received another $1.35 billion investment from SoftBank, a Japanese venture capital firm whose Vision Fund specializes in disruptive technologies. SoftBank had already ponied up $900 million in capital to Cruise in 2018, making it a major backer of the company alongside GM, Honda, Microsoft, and Walmart.
“This additional capital will help us expand our world-class team and quickly scale this technology across San Francisco and into more communities,” said Vogt.
Tesla FSD is the “wrong thing at the worst time”
By comparison, Tesla has deviated from the standard approach of building a dedicated robotaxi fleet. It argues lidar is an unnecessary crutch, since cameras costing a fraction of the price are sufficient.
As a result, Musk has already sold his full self-driving feature, or FSD, to tens of thousands of Tesla customers before completing development. While risky and unproven, in theory it is the fastest way to scale up the technology. All that these Teslas would require to be transformed into autonomous cars is a remote software update.
Musk recently hiked the price of FSD by a fifth to $12,000 and pledged last week to solve FSD this year in lieu of launching new products. Currently 60,000 Tesla drivers have been granted access to its latest test version of FSD.
Experts are skeptical, however, as the cameras, for one, can be blinded in certain lighting conditions. Tesla’s FSD feature also currently requires constant human supervision at all times, since the carmaker warns users that its software may cause the vehicle to “do the wrong thing at the worst time.”
Contrary to Musk’s typical brand of overambitious bravado, Cruise’s Vogt said he was well aware of the risks.
“We’re being cautious and careful with this rollout,” he posted to Twitter. “This has never been done before in a major U.S. city, so we’ll expand the service area, hours, etc. as we learn what works and what doesn’t.”
Never miss a story: Follow your favorite topics and authors to get a personalized email with the journalism that matters most to you.