A Robotaxi Regulation Wishlist
The decisions San Francisco and California make now will have a tremendous precedent-setting impact.
Robotaxis have been officially unleashed upon San Francisco.
This could be a pivotal moment in the history of urban transportation. The decisions San Francisco and California officials make now will have a tremendous precedent-setting impact. They could help determine whether robotaxis compliment existing transportation services and benefit cities; or whether they “ream out cities anew” in a second auto age, as urban planner Jeff Speck warns.
Here’s where things stand: Despite protests from San Francisco transportation and public safety officials, the California Public Utilities Commission yesterday approved Cruise and Waymo for 24 hour commercial robotaxi service across virtually all of San Francisco. There are still plenty of balls in the air. The CPUC has vowed to revisit its robotaxi rules, perhaps as early as November. San Francisco officials are discussing their options, including a possible appeal of the CPUC’s decision. The robotaxi companies themselves have said they will thoughtfully expand their services, though there are no limits on the number of vehicles they can deploy.
Time will tell how these robotaxis perform now that they’re being commercially deployed at scale. The early days will probably be rocky. But my bet, after covering robotaxis for over a year, is that these vehicles’ most disruptive driving behaviors can be significantly improved. They can learn to better recognize and respond to emergency vehicles and construction sites. They can learn to pull over in challenging scenarios, rather than idle in the middle of the street.
The bigger challenges will be political, economic, and environmental. How can San Francisco avoid massive increases in traffic? What will happen to the city’s taxi and Uber drivers? How will public transit ridership and revenues be impacted? Or the city’s broader efforts to get people out of cars?
In a previous post, I discussed why outright robotaxi bans are naive at this point. The more realistic option is to evaluate the potential negative impacts of robotaxis and design policy remedies specifically tailored to those impacts. In the process, I argued, there’s a window of opportunity to advance elusive policy goals that could actually make cities more humane and less car-dependent.
At this pivotal moment, it would be valuable to start brainstorming specific robotaxi regulations. As the recent CPUC approval process demonstrated, the legal and administrative frameworks that govern robotaxis are quite complex. Certain regulations would require new CPUC rules. In California, any policy that could be interpreted as a new tax requires a vote of the people. The state legislature could also write new laws regulating robotaxis. That would require San Francisco’s usually innovative and proactive Sacramento delegation to wade into the robotaxi debate — something they have avoided thus far.
Instead of calling for outright bans, San Francisco officials could orient their lobbying efforts towards shaping future CPUC rules in ways that benefit the city. They could lay the groundwork for a ballot initiative levying special taxes on robotaxis. They could redouble their efforts to carve out street space for transit, people, and bikes, in anticipation of the robotaxi deluge.
In a way, robotaxis provide the opportunity to correct the mistakes that were made with the emergence of Uber and Lyft. Robotaxi regulations could serve as pilot projects that get the public used to policies that could someday be rolled out to all ride-hail vehicles, or even all cars.
So here is a robotaxi regulation wishlist, based on urban planing best practices, for your consideration. I’d love to hear your thoughts in the comments.
Robotaxi congestion pricing: Taking a robotaxi to downtown San Francisco at 9am? Prepare to pay up. To avoid a traffic nightmare, robotaxis need to be disincentivized from traveling through the busiest parts of town at the busiest times, especially when there are so many alternatives available. Congestion pricing fees could go toward public transit or a taxi medallion bailout fund. While a broad based robotaxi surcharge — like San Francisco’s modest, existing ride-hail tax — would be simpler to implement, it wouldn’t provide congestion relief when and where it’s needed.
Robotaxi-free corridors: Major bike, transit, and commercial corridors should be off-limits to robotaxis. Valencia, Polk, Haight, and Irving immediately come to mind. Robotaxis could serve these destinations by picking up and dropping off at cross streets.
Designated pickup and dropoff spots: In busy parts of town, every block should have one or two designated pickup and drop off spots that robotaxis are required to use. This would cut down on the scourge of double parked cars. An easy place to put these spots would be next to intersections, where parking spaces are already being removed for pedestrian “daylighting” projects. Robotaxis should also be kept at a distance from major events, like Outside Lands and Giants games.
Robust data sharing: Cities should have a better accounting of disruptive incidents caused by robotaxis. But they should also get a sense of robotaxi travel and demand patterns. This information could be used to shape transit service, or augment the congestion pricing district.
Disability access: A certain share of robotaxis on the road should be wheelchair accessible, just like regular taxis. I’ll note that disability access comes in conflict with other goals on this wishlist, like banishing robotaxis from certain streets. These are some of the tough tradeoffs that will have to be weighed in a robotaxi future.
It’s obviously very easy for me, a freelance journalist and blogger, to jot down ideas on a screen. But I think this is a very important moment in a very important policy conversation that has never received the attention it deserves. Decision makers are not in a position to have a blue sky brainstorming session in public. Those of us with that liberty have a role to play in sparking a broader conversation about how to navigate this brave new world.
great read, as always. One addition to your list might be something to address the network scalability of these services and how that relates to “Pool”/“Line” features. It always felt sad to see two people waiting at the airport pre pandemic for their rides knowing that they may well be headed for a destination a block apart, and yet their rides fail to be matched not because of an algorithmic failure but rather the deliberate walled garden between the person ordering Uber Pool and Lyft Line. And now, in the post-pandemic era, it seems the companies have been really sluggish to launch them again at all, and I hypothesize that the economies of scale and tipping points might be what is holding them back from offering a compelling service at lower price than standard single person rides.
In a more mature state, robotaxi service feels more like something that aught to be consolidated at a regional level akin to a utility or municipal service. By having a regulated monopoly, perhaps competed for every 5 years, you could probably negotiate better policy terms with the operators and also reasonably expect a higher rate of shared passengers per vehicle by ensuring that all robotaxi requests are hailed through the same system. That data on when/where matching is most abundant also seems like an unbelievably good data source with which to plan bus service - if you can put terms for having access to it.
This is regulated monopoly model basically akin to how public transit works in Europe and in some US cities where private operators bid on service contracts for a municipal public transit system. Also true for paratransit services which are more similar. It’s also more similar to taxis - where within a given location there’s often only one operating taxi service but with tight municipal control.
Anyway, I don’t really trust people to get that kind of contract language right to mandate a single robotaxi operator and procure it today, but I hope we can get there.
I don't get the argument for treating these differently from any preexisting taxis or ride hailing. The externalities seem the same