Whether robotaxis will really prove cheaper than buying your own car depends heavily on where you live — and how much inconvenience you’re willing to put up with.
Self-driving cabs have been touted as a panacea to the drawbacks of conventional vehicle ownership, from environmental pollution to congestion to highway crashes. The Trump administration has championed autonomous vehicle technology, and companies from electric carmaker Tesla to Google’s Waymo and General Motors have invested millions of dollars.
Achieving society’s ambitious goals for so-called self-driving cabs requires “cost competitiveness with conventionally driven vehicles,” Ashley Nunes, a researcher at the Massachusetts Institute of Technology, wrote in a study published in April. “This may, all else being equal, be a difficult proposition.”
Using San Francisco, the headquarters of ride-sharing firms Uber and Lyft, as a test market, Nunes and co-author Kristen Hernandez found that robotaxis with costs and operating conditions comparable to those of existing cabs would be more expensive than owning a conventional vehicle. The study estimates fares of $1.58 to $6.01 a mile, using a car minimally priced at $15,000, compared with personal ownership costs of $1.61 a mile.
The difference is primarily fueled not by the typically higher cost of buying a self-driving car but by inefficient utilization of robotaxi fleets, presuming operators relied on the common practice of carrying one passenger per trip.
[Related: Forget self-driving cars; Boeing just tested a self-flying taxi]
“People think this is going to be some sort of cash cow,” but even generous calculations don’t back that up, Nunes told the Washington Examiner. “When we did this study, we did not expect to find this result.”
To offer fares comparable to the cost of owning a vehicle, while keeping the typical profit margin for a conventional cab company, would require boosting robotaxi occupancy by 30%.
In real-life terms, “you need close to three people in the car,” Nunes said. “This is not just you, your mother, and daughter. This is three separate fare-paying passengers.”
To make the business model work with only one fare per trip would require a robotaxi to have a passenger for every mile it travels, compared with current averages of about half that.
It’s a problematic target. Meeting passenger demand requires taxi operators to have enough vehicles available for peak travel periods, such as morning and afternoon commuting hours. That necessarily means that some of those vehicles will be idle outside of those hours, Nunes explained.
[Also read: Flying cars coming in five years, says Intel’s drone developer]
“By default, you’re going to have dead time for your assets,” he said. That matters more for a commercial vehicle, which is usually still operating even when it has no passengers, and thus accumulating miles that wear down parts as well as consuming fuel, than it does for a individual’s automobile.
“People say personal cars are poorly utilized,” being driven only about 5% of the time, Nunes said. “That’s true in terms of the time the car spends being physically driven around. That’s not true in terms of miles. Every single trip you take in a personal car serves a purpose. This philosophy effectively changes in a robocab situation: Not every trip is generating revenue. Only about half the trips are generating revenue.”
Another wrinkle is that making a single-passenger model, which most passengers prefer, work with robotaxis would require operators to settle for a profit margin of 17 cents a mile rather than the 27 cents that’s common now, he added.
Marketing shared trips is difficult since passengers are loath to pay for a ride that includes an unknown destination and travel time that may be significantly longer. Without a substantial discount, taking public transit such as buses or subways may become a more practical option.
Other variables include salaries for fleet monitors and the considerably higher costs of providing cabs in less populated rural areas with longer trips and fewer passengers. The San Francisco market examined by Nunes and the New York City market evaluated in a separate study by Swiss lender UBS are dense urban areas.
How these findings will affect the nascent industry remains to be seen. Despite the cash funneled into both self-driving traditional cars as well as taxis, Congress has yet to develop any kind of regulatory system that would give automakers confidence to mass-produce such vehicles for nationwide use.
For now, the vehicles are governed by a patchwork of state laws and regulations, despite support from both the Trump and Obama administrations. In December, Transportation Secretary Elaine Chao outlined a broad approach to preparing for the technology’s implementation that included revising existing regulations and focusing on safety after two widely publicized deaths involving autonomous cars in 2018.
“If developed safely and responsibly, these automated driving system technologies could potentially save lives,” she said at the time. “Historically, human error has been a factor in 94% of fatal crashes, which automated technology could help address.”
Tesla founder and CEO Elon Musk, whose company built the vehicle in one of the 2018 accidents, expects to begin offering robotaxis as early as 2020.
He plans to use a model similar to Airbnb or Uber that lets self-driving Tesla owners add their vehicle to the network through a company app, generating a profit of about $30,000 a year per car.
“Next year for sure, we’ll have over a million robotaxis on the road,” he told investors in late April. “The fleet wakes up with an over-the-air update. That’s all it takes.”
Initially, robotaxis will be equipped with the steering wheel and brake and accelerator pedals required for a user to take over if necessary, Musk said. “Then, once regulators are comfortable with us not having a steering wheel, we’ll just delete that. And for cars that are on the fleet, obviously with the permission of the owner if it’s owned by somebody else, we would just take the steering wheel off and put a cap where the steering wheel currently attaches.”
Waymo, the self-driving vehicle division of Google owner Alphabet Inc., is still testing a fleet of robotaxis in Phoenix, using technology under development for more than a decade. The company has offered scant detail about its financial performance, grouping the business in a larger unit that lost $868 million in the first three months of 2019.
Developing the ride-sharing business is a priority for Waymo, Chief Financial Officer Ruth Porat said in late April, and “we’re pleased with the expansion in the number of Waymo riders that we’ve added since the last quarter.”
Earlier this month, the Japanese conglomerate SoftBank invested another $1.15 billion in Cruise, the GM unit developing a fleet of robotaxis, bringing its total valuation to $19 billion.
“Developing and deploying self-driving vehicles at massive scale is the engineering challenge of our generation,” Cruise CEO Dan Ammann said at the time. “Having deep resources to draw on as we pursue our mission is a critical competitive advantage.”
