San Francisco was supposed to have a vote the other day to expand Cruise and Waymo robotaxis services in the city. The problem is, some prominent agencies in the community claim that robotaxis services in San Francisco have been creating a number of problems. Others disagree. The vote that was supposed to take place a few days ago has been moved to July 13 amidst the opposition to the expansion and the opposition to the opposition.
This vote isn’t just about a step-change expansion in services. It’s on whether or not to provide these companies with the opportunity for unlimited robotaxi expansion — 24 hours a day across the city. So, let’s take a look at what the issues are that are stalling the robotaxi revolution in its birthplace.
Robotaxis Still Learning, Creating Too Many Incidents On Road
One problem is reportedly that robotaxis are involved in a lot of incidents that aren’t necessarily major, life-threatening issues, but are demanding too much time from municipal emergency vehicles. “According to Assistant Deputy Chief Darius Luttropp of the SF Fire Department (SFFD), an increasing number of non-life-threatening incidents involving self-driving cars are hindering their ability to respond effectively,” CBS News writes.
“These cars are on the streets learning … it’s like these cars are in school right now, and we don’t feel like they’re quite ready to expand to the level they want to expand. We’ve asked to get more insight into the process of their operations,” stated Luttropp.
The San Francisco Municipal Transportation Agency (SFMTA) indicates that, particularly in recent months, the number of incidents involving these robotaxis has jumped. The agency says the number of incidents involving Waymo and Cruise vehicles almost tripled compared to the number in the previous two months. As a result, San Francisco Municipal Transportation Agency wants incremental expansion, rather than 24-hour, citywide expansion right now. Head of the agency, Jeffrey Tumlin, has said that these robotaxis have only “met the requirements for a learner’s permit.”
Jalopnik has one story from last week about this lack of experience or understanding. “On Sunday, San Francisco also had a Pride Parade and the Giants played some baseball, which happens almost every day. Two Waymo robotaxis didn’t seem to know how to manage the resulting traffic and responded by stopping dead in the middle of intersections. The incidents were covered in an ABC7 Bay Area segment and on NBC Bay Area on Monday evening, showing scenes of the inaction.” Yikes. San Francisco Fire Chief Jeanine Nicholson adds that robotaxis have “driven into shooting scenes and fire scenes and just been a menace in so many ways to the San Francisco Fire Department, our trucks, and ambulances.” To be honest, that doesn’t sound great. They’ve also entered active construction sites and crime scenes. I’m starting to sense a problem.
Also, Jalopnik reports that the SFMTA’s letter opposing robotaxi expansion “referred to data that the injury collision rate of Cruise’s cars in particular was six times higher between June and November of last year compared to the 2021 national average for human drivers.”
“Are Incidents Really Abnormally High? Show Us The Proof”
Not everyone agrees that the robotaxis are creating an abnormal number of incidents, though. In fact, the California Public Utilities Commission (CPUC), which was supposed to take the vote on Wednesday, published a 23-page report on the topic in place of the vote. It’s not thrilled with the critics’ data collection and analysis methods for determining (or assuming) a higher rate of road incidents among robotaxis. “San Francisco’s comments include its own analysis of Waymo’s safety record, based on data available from NHTSA, the California DMV, and the Commission,” the report notes on pages 18 and 19. “San Francisco states that its analysis indicates the Waymo AV’s injury collision rate appears to be higher than average human drivers. However, we find San Francisco’s analysis lacks sufficient rigor and nuance to form a basis for modifying the Resolution. It highlights the need for enhanced systematic data collection that supports objective analysis of AV performance. We encourage San Francisco, along with all parties, to participate in the continued development of AV data reporting requirements through the rulemaking process. We discuss below our concerns with San Francisco’s conclusions, including its statistical methods for assessing the frequency of collisions and the lack of contextual awareness in assessing responsibility of the collisions cited.
“Regarding the frequency of collisions, San Francisco’s analysis necessarily covers a very limited data set – 6 months of operation and an estimated 1.9 million vehicle miles traveled (VMT) – due to the nascent nature and small scale of Waymo’s AV operations relative to conventional human-driven vehicles, which constitute multiple orders of magnitude more VMT. Extrapolating from less than 2 million miles to 100 million, and then comparing to a national average without normalizing for factors such as roadway type (e.g. arterial vs. local street) or land use context (e.g., urban, suburban, or rural) introduces an unacceptably high degree of statistical error and uncertainty.
“Regarding collision responsibility, San Francisco’s analysis appears to omit or overlook relevant facts present in the data and collision narratives that are critical for understanding the context of the cited incidents. The examples of two injury collisions upon which it seems San Francisco bases its analysis of Waymo’s relative injury collision rate (included below in Appendix A as entries for June 2022 and July 2022) are problematic in this regard. According to Waymo’s account as submitted to NHTSA, the June 2022 collision does not appear to involve any contact with the Waymo AV. The narrative of the July 2022 collision indicates the Waymo AV was rear-ended by another vehicle, which immediately left the scene. Note that no determination of fault, of the AV or otherwise, is evident through these reports. The highest reported injury severity of these collisions was minor. While we acknowledge the need to proactively evaluate early data and less severe collisions as leading indicators of safety performance in hopes that such proactive monitoring will help prevent additional collisions and/or more severe incidents, the shortcomings of this analysis again highlight the need for systematic data collection that supports objective analysis of AV performance.”
So, it’s not a full rebuttal arguing that robotaxis are definitely not involved in more incidents, or even involved in fewer incidents. It’s just that the CPUC doesn’t see the critics as performing a particularly good statistical analysis and doesn’t assume their conclusions are correct.
Waymo and Cruise also wrote their own retorts. “We’re proud of our safety record, which is publicly reported and includes millions of miles driven in an extremely complex urban environment,” said Hannah Lindow, a Cruise spokesperson. She added that various businesses and disability rights groups also sent letters in support of the robotaxi expansion.
Robotaxis Blocking Emergency Vehicles?
Another claim put forth was that robotaxis have been blocking emergency access vehicles somewhat frequently, that they are not interacting appropriately with emergency vehicles. Again, the CPUC wants to see proof.
“We acknowledge San Francisco’s comments and agree, as we have expressed earlier in this resolution, that incidents such as unplanned stops and improper interactions with first responders are concerning and represent hazards to passenger and public safety. We appreciate San Francisco’s efforts to share information on incidents it becomes aware of and are requesting party comment on how to formalize such a process. However, these anecdotes do not represent a sufficiently robust set of facts upon which to alter the Draft Resolution’s findings or conclusions. The Commission has initiated a process to update data collection requirements in the AV program through R.12-12-011 and encourages San Francisco to participate so that rigorous, non-anecdotal incident and other AV operations data may be systematically collected, analyzed, and acted upon in the future. At this time, the information shared by San Francisco does not alter our conclusion that Waymo’s advice letter meets the requirements of the Deployment Decision and that its PSP is complete and reasonable per existing requirements. Any future modifications to these requirements or the standard of review are more appropriately addressed through the rulemaking, not the advice letter process.”
Taxi Drivers Protesting
Tax drivers have also reportedly been protesting, to some degree. “Barry Taranto, a cab driver of 21 years, is glad for the delay. His main concern at this time is the technology causing problems for city streets; ultimately, he wants the state to better adapt laws to the times. He sees several robotaxis each night disrupting other traffic,” The San Francisco Standard writes. Well, is robotaxis causing traffic problems really his main concern? Or is his main concern loss of business? Those are rhetorical questions, but Taranto disagrees with my assumption. “My main concern now is the technology is not advanced enough for them to operate properly on the streets,” says Taranto, a member of the San Francisco Taxi Workers Alliance board of directors. “The cars get confused and results in them being stalled out or in the way.”
And, let’s be honest, that’s a legitimate concern — even more so if you consider the following point. “Marcelo Fonseca, a cab driver of 34 years, is primarily concerned by a proposal that lets another new technology company operate rides in San Francisco without $250,000 medallions that several taxi drivers are still paying loans on.”
Indeed — why should taxi drivers have to spend $250,000 on taxi medallions while Cruise and Waymo robotaxis drive people around without them? That’s a matter I don’t see discussed much.
I don’t like paywalls. You don’t like paywalls. Who likes paywalls? Here at CleanTechnica, we implemented a limited paywall for a while, but it always felt wrong — and it was always tough to decide what we should put behind there. In theory, your most exclusive and best content goes behind a paywall. But then fewer people read it! We just don’t like paywalls, and so we’ve decided to ditch ours.
Unfortunately, the media business is still a tough, cut-throat business with tiny margins. It’s a never-ending Olympic challenge to stay above water or even perhaps — gasp — grow. So …