Lambert here: As I keep saying, we will need to optimize the driving environment for robot cars, not human drivers (for example, put first responders in robot helicopters), and make sure the robot car fleet owners are immunized from lawsuits. The assumption here seems to be that robot cars scale (including EV robot cars). Maybe when the streets are emptied of everything but robots, they will. Meanwhile, I will sit back and wait for the time all traffic in the city of San Francisco is bricked by a some thirteen-year-old in a Palo Alto garage [snarl].
By Wolf Richter, editor of Wolf Street. Originally published at Wolf Street.
Robotaxi regulator, the California Public Utilities Commission, after hours of testimony from supporters and opponents, voted on Thursday to allow GM’s Cruise and Alphabet’s Waymo to charge riders for driverless robotaxi service, day and night, anywhere in San Francisco, with no cap on fleet sizes. They can now commence full commercialization of robotaxis in San Francisco.
We’ve been seeing them everywhere in San Francisco: Fully autonomous vehicles (AVs) from Cruise and Waymo with no driver and one or two people in the back, and vehicles with no one in them at all. They’re generally well-behaved. They smoothly roll up to a stop sign, come to a complete stop, stay there for a couple of seconds, and then softly accelerate away. They stop when the light turns yellow and don’t floor the accelerator to get through the intersection on dark-yellow or whatever. And they don’t do donuts in intersections.
San Francisco is a challenging environment to drive in, immensely congested, with steep hills that impede visibility at intersections, lots of pedestrians and bicyclists, and people on weird electric conveyances with one wheel or two wheels or three wheels.
Occasionally, a hilarious video goes viral of a police officer or firefighter trying to tell a robotaxi what to do. Someone came up with the prank of the year, or whatever, by getting a bunch of empty robotaxis to all meet in one intersection, clog it up, stop, and bring everything to a halt for hours until human drivers could sort it out.
Human drivers do stupid things all the time, from donuts in intersections to going down embankments. They injure and kill pedestrians, bicyclists, and each other and their passengers. And they constantly get into minor accidents that no one even tracks. That’s normal. But robotaxis are held to a higher standard. And they’re doing amazingly well in that regard.
But they do some things humans don’t do, like completely blowing off instructions from first responders and just freezing in place.
So this show has been going on in San Francisco for a while, to the great amusement and frustration of everyone around. But the robotaxi companies have been limited in the commercialization of the service. Waymo was allowed to only offer free driverless rides (to charge, a safety driver had to be in the vehicle); and Cruise was allowed to charge only for rides at night in limited parts of the City; the rest of the time, it could offer only free rides. But those restrictions were lifted on Thursday with the vote of the CPUC.
City officials, the fire department, and the police department have for weeks urged the CPUC to slow down the rollout of full commercialization because they worried about the interference of robotaxis with “the work of first responders,” as Fire Department Chief Jeanine Nicholson told the commission during the public hearings on Monday. “Our folks cannot be paying attention to an autonomous vehicle when we’ve got ladders to throw,” she said.
Officials cited 55 incidents over the last six months where robotaxis got in the way of first responders.
Through June, City agencies – which last year began collecting data on disruptive robotaxi incidents to prove that robotaxis weren’t ready for full commercialization – tallied 600 such incidents, such as interfering with public transportation or blocking traffic. City officials that have to deal with these messes worry that full commercialization of much bigger fleets would increase those disruptions.
Other opponents, including activists and unions, worry about the jobs of drivers – the jobs of lots of drivers. And you can see where this is going by what GM said in its Q2 earnings call, via Cruise CEO Kyle Vogt:
“There’s over 10,000 human ride-hail drivers in San Francisco, potentially, much more than that, depending on how you count it. Those drivers of course aren’t working 20 hours a day, like a robotaxi could. So, it does not make a very, high number to generate significant revenue in a city like San Francisco, but certainly, there’s capacity to absorb several thousand per city at minimum.”
And people worry that large fleets of robotaxis will make congestion even worse, and they worry about all sorts of other mayhem.
Waymo said it has a permit for 250 AVs and deploys about 100 at any given time. Cruise said it operates 100 cars in San Francisco during the day and 300 at night. So the current fleets are relatively small. But with full commercialization, the fleets are bound to get much larger.
Both companies have invested many billions of dollars in developing this technology, and they will want to eventually get a return on their investment. So ramping up the fleet size would be the first step. But that also is a huge capital investment because the AV technology, for now, makes those vehicles expensive.
Cruise CEO Vogt addressed some of the issues of ramping up the fleet size:
“As for what it would take to blanket a city like San Francisco, our goal is – I think I’ve said on previous calls – is to make sure as we ramp-up manufacturing capacity. We’ve got a variety of markets to absorb those vehicles. And there are practical reasons to ramp-up gradually in a city … as it’s transitioning to a new form of mobility. So, it’s not our intention to sort of produce vehicles and sort of direct them all into a single city.”
Vogt also said that Cruise exceeded 3 million miles in 49 days in the cities it operates in; that it’s now doing over 10,000 rides per week; and that its rides are growing at 49% per month on average over the last six months.
Investors can be an impatient bunch, and analysts are poking around during earnings calls. After sinking billions of dollars into robotaxis, these companies are under pressure to show significant revenues. Uber and Lyft got lots of revenues, but they lost eyewatering amounts of money year after year – in part due to the cost of human drivers and related expenses.
Cruise and Waymo are now replacing the cost of drivers with other costs, including the much higher costs of the vehicles and the much higher costs of the people who are building, expanding, and maintaining the technology. So that equation isn’t going to be easy to work out.
Meanwhile, we cannot wait for this to be truly commercialized to an every-day-for-everyone level to where it’s cheaper than car ownership, so we can get rid of our car that is mostly parked somewhere, and get rid of all the hassles and costs, and just let the machines do the driving when the driving needs to be done. We’ve already waited for a decade; what’s another decade?
37 comments
Comments are closed.
Autonomous semi-trucks when? Then the real job carnage starts!
Wasn’t there a Stephen King movie about that?
Can’t believe with all of the problems of them blocking emergency vehicles that this went through. But then again I can.
There is also going to be a pricing issue. Uber grew not just based on ease of hailing, but that it was cheaper than a city regulated taxi much of the time. It may still be on people’s apps, but at least here in NYC, I don’t begin to see the number of car pickups I used to see. Not all of that is a result of work from home. They are also no longer cheaper.
These driverless vehicles have repeatedly been roaming the city empty when they are free. If they need to start showing ROI how long do they really have for the ubiquitous nature of a full rollout to overcome natural hesitancy on the part of the public. And can they do that charging rates that are equal to or greater than fully manned vehicles, something that may be necessary considering the expense of production and maintenance.
And the lawsuits, the probable public activism/vandalism, and bad publicity sure to be part of any event that includes death or destruction resulting from interference with a first responder all are going to put increasing pressure on robotaxi companies. I may also await the kid in the basement causing havoc just to see if they can, but even without that the obstacles these companies face are going to be much more formidable than They believe.
> There is also going to be a pricing issue.
Enshitification will take care of that. How?
Here is MIT guy explaining the true reason behind autonomous vehicles. Oldie but goodie.
Object to ads being blasted in your face, the ride price doubles. Want to listen to music, that will be extra, the list of creepazoid things silicon valley dreams up to extract moar from customers is endless.
Never touched on by either Wolf or any commenters is the pecuniary interests of the CPUC (I pronounce it ‘see puke’) as in do they hold stawks in either of those companies or are they or their relatives gorging on direct payment from Google or GM? It must be asked by someone as corruption is now blatant everywhere one cares to look.
The techno-fetishism is off the charts. Their lives would be perfect had they been born as a silicon chip.
I understand the point about underutilized cars and mis-allocated capital of all the parked vehicles around the world.
But is it not the ones in motion that frustrate?
There are simply too many of us… me included. And I hope Lambert was being sarcastically facetious abut protection from lawsuits!
I nearly got hit 3X yesterday on my 8 mile round trip bike ride to play disc golf. Sunday, mid-day.
As bad a ride as I have had in a while, in terms of near-miss altercations of David v Goliath.
Two of the drivers were viewing their gizmo’s, the third followed three young men on bikes with 2 cycle weed eater conversions through a stop sign.
( It is the season of The Touron, and we will never solve Young Men).
It seems that it is the rule, not exception, that folks are on and looking at and fiddlng with their Gizmo while in control and command of their vehicle. And I can’t help but assume some of these folks will be behind the technology and programming of autonomous vehicles.
What could possibly go right?
I ride large city buses almost every day. I have gotten used to near accidents as there is something egregious four maybe five times a week. Buses are not small, but apparently many people in cars or yes on bikes cannot see them and almost hit them period, or see them and cannot evaluate when they will pass them or the speed the bus is going in order to safely change lanes in front of them. Luckily the buses were able to stop short of running into them.
It isn’t just gadgets distracting people on vehicles. Large numbers of drivers and riders just aren’t focused on what they are doing anymore.
And to bring this back to autonomous vehicles, I don’t think being surrounded by careless unaware human operators is good for them either.
And then there is this
https://www.latimes.com/california/story/2023-08-12/cruise-robotaxis-come-to-a-standstill
I read about that earlier today. It was chaos. You have to wonder what would have happened it this had taken place on the Golden gate bridge during rush hour. Governor Newsom apparently lives in a $3.7 million mansion in Fair Oaks in in Sacramento County. I wonder what would happen if some hacker sent a coupla dozen of these robot cars to block his street. Would the media actually cover that or just look the other way?
I read Wolf’s original piece on his web site the other day and then noted separately the LA Times piece about the robotaxi-jam (“robojam”?). I hadn’t considered The Rev Kev’s idea that they could be, as it were, strategically deployed!
If anyone needs his or her life disrupted, Newsom would be near the top of my list. He and his CA colleagues have destroyed one of the most beautiful cities in the world. I lived there off and on around 1980 and then later in the late 1990s whilst working in Silicon Valley and now have zero desire to return. Sigh.
Reminder: Every member of the California Public Utilities Commission has been appointed by The Newscumbag.
Apparently placing a traffic cone on the hood of these things will bring them to a halt. Not that I am recommending inconveniencing our corporate masters and their exploitation of public infrastructure extract profit and socialize costs and risks. .
I have been wondering how soon high-visibility vests worn by road construction workers, cyclists, and school crosswalk attendants will be emblazoned with a quality rendition of an orange traffic cone. Imagine a pedestrian being able to cross a congested AI-only roadway (you know those are coming) with impunity by simply wearing one of these. Until, of course, the software engineers disable the recognition of all traffic cones as a driving safety signal / device. Hilarity ensues.
Do whatever it takes to save jobs, to prevent the dehumanization of our cities and give a middle finger to Silicon Valley.
As an ex-taxi driver, and my wife being nearly clipped on a bike by an an autonymous taxi–boy, it does feels ridiculous riding after it and shouting at it,
I have vowed to do whatever I can to smash and hinder such vehicles, only when empty.
If people in it, might it be considered an assault on them?
The cameras and sensors are the weak spots. Shoot with a supersoaker squirt gun with 10% white water based paint, or directly spray paint them, scratch the lens, slash the tires when sitting still. Pour old motor oil inside after breaking the window. Do it now before they become legion.
That’s one incident every three days. How long before somebody dies in an ambulance or an apartment building burns down?
Then GDP will be increased by the payment to the undertaker and the contruction cost of replacing the apartment building. It’s a win! The disasters wrought by climate change will make us all rich!
I would like to see comparable statistics for “drivered” vehicles blocking first responders. Many, many times I have seen drivers not pulling over for ambulances and fire trucks, or worse.
worth watching the history of transportation in SF https://www.pbs.org/video/moving-san-francisco-dqxmfa/
Wolf’s story doesn’t address any of the important issues here.
While the huge hypothetical potential of every previous major breakthrough in transportation technology (RRs, waterways, highways, aviation, container ports, etc) was always clear they took a very long time to mature and a longer time to displace earlier technologies. Early incremental phases identified lots of problems and allowed the path of technological and commercial development to shift. Private sector players were diverse and highly competitive but depended on massive public/governmental support and oversight. The provision of core infrastructure and operating capacity were separate. Commercial expansion depended on huge scale economy/service advantages over incumbents.
AVs have been following a completely different path. Investors have been focused on the ideal future potential—backers talk about a future Jetsons-like world where cars and accidents have vanished–and and haven’t laid out any viable intermediate stages. Absolutely no one can explain how a system with greatly expanded AVs coexisting with cars would work—who owns/maintains/insures the vehicles, who builds the communication systems and keeps it secure, who makes needed changes in roadways, etc. etc. Absolutely no one can lay out a path to viable commercial operation, where revenues cover costs and fund ongoing growth. Absoluely no one can document that pre-Jetsons AV operations would actually achieve major cost/efficiency advantages or actually achieve major reductions in accidents and fatalities. While it would be hypothetically possible for a Federal oversight body to drive the same huge benefits that the public sector provided for highways, ports and aviation, that clearly won’t happen, and none of the major investors in AV want it to happen.
Unfortunately, AV development is following a Venture Capital model, where a very small number of players with access to massive pools of speculative capital are pursuing Powerball-jackpot type levels of equity appreciation. Past transport technologies were explicitly designed to produce massive public benefits over long time periods. AV development is based on the irrational faith that any problem can be solved by VCs who will focus on allowing early investors to cash out with big profits regardless of whether the technology actually works or whether a commercially substaintiable industry actually develops. Instead of a diverse, competitive private sector role (lots of railroads, carmakers, airlines, shipping companies, equipment manufacturers, terminal operators, etc) all we have is Google and General Motors. Why would anyone think that these companies could be trusted to produce the huge promised public benefits?
As the California PUC decision shows, it is child’s play for these investors to capture whatever limited protections for the public (and long term AV industry) interests might exist, and overwhelm public “debate” with cheap PR narratives about Luddite level hatred of technology and the imminent promise of eliminating all traffic fatalities.
There is little discussion as to why most recent investments into a glorious future where AVs replace cars have failed miserably. Having sunk so much into AVs, Google and GM have apparently chosen to double down on hype, since an open admission that they have been burning the shareholder’s cash would be an admission that management never had a coherent plan, and would undermine their ability to use manufactured narratives to maintain inflated equity values.
What will it mean for Uber’s stock price that GM and Google claim they are about to take over its entire market?
I’ve not seen one of these contraptions in operation, but I imagine they have cameras and sensors that give them travel orientation (including your photo as you enter the car). I can also imagine, as a former SF native, that there are pirates that will disable these sensors and make the vehicles useless (or worse). And don’t expect the police to investigate these transgressions—the police hate them. too!
When I lived in SF (long ago) taking the electric Metro was the best way around the community. Now these individual cars are Waymo better? While double-parking in the City has always been an art form, these robot cars are likely rank rookies.
Evidently the recent robo-car failure was blamed on GG Park concert goers overwhelming the cell phone network that provides the robo-cars with their driving data. Fail-safe for sure!
I used to live in San Francisco, not going to say over what period, but I left a year or two after Covid. Anyway, right at the time I left, they were still working on the Central Subway line which would serve areas such as Chinatown, and till now there are still many areas in San Francisco that are not served by the Metro, like the whole Geary Boulevard corridor, Presidio, etc, although there are buses serving those areas. Anyone who’s ever taken the 38/38L in the morning to go to work will not say that it’s a pleasant experience.
I am not saying I am in favor of these robo taxis, but the transportation system in San Francisco was far from decent. Metro stations and bus stops in San Francisco were equipped with Next Bus equipments displaying how long you have to wait for a particular Metro train/bus, and it was super frustrating how inaccurate they often were. I am someone who advocates taking public transit if possible, I didn’t even have a car, but at the end of my tour of duty there, I had given up on the whole thing ever improving.
https://sf.curbed.com/2019/9/16/20868965/sfmta-muni-rail-on-time-august-2019-late-very-late
I moved around when I was living in SF, but for a number of years I had to take the N Judah to work, and man that was horrible. What’s funny is that a local also has had a less than stellar experience taking the N Judah so he created a blog documenting his experiences, https://njudahchronicles.com.
There has been a problem in San Francisco, in the past decades of not enough taxis, but other then that, they were just fine when I used them. Any problems could have been solved by adding more taxis. This is just another grift.
Except that adding more taxis undermines the financial viability (including driver income) of existing taxis (and Uber/Lyft), while adding to congestion. There’s (pun noticed) no way around lots of vehicles on the street.
Yes, driving around San Francisco has always been an adventure. Adding more cars does makes it worse, but if I had to choose, the regulated taxi service with their human drivers is what I would choose to expand. The robotaxis are the worse of both worlds.
R.U.R.al Delivery
I don’t see how this system is going to be viable in a suburban/rural environment. The suburban commuter will find it prohibitively expensive, and the suburban household will not use them for every little errand. If it isn’t more expensive than current train commuting then it’s highly subsidized.
The car is the last vestage of “freedom of movement” in the US. I can travel anywhere in the lower 48 at anytime. Yes I know it is a privilege, because of the costs, but it’s there. If the automotive goes we truly become a dependent population.
> Meanwhile, I will sit back and wait for the time all traffic in the city of San Francisco is bricked by a some thirteen-year-old in a Palo Alto garage
LOL … #Exactly
If you’re going to San Francisco
Be sure to have your robo-taxi fare
If you’re going to San Francisco
You’re gonna need to tell it your destination there
For those who come to San Francisco
Any old time with be an automation-in there
In the streets of San Francisco
Robo-taxis snatching cabbies fares
All across the nation such a strange vibration
People in motion
There’s a whole generation with a new explanation
People in motion people in motion
If you come to San Francisco
Be sure to have your robo-taxi fare
If you come to San Francisco
You’re gonna need to tell it your destination there
San Francisco, by Scott Mckenzie
https://www.youtube.com/watch?v=7I0vkKy504U
Mobile Port o Sans.
This is what will wipe out the investors. No, really.
Robotaxis will ultimately appeal not only to people who need a lift but also as summonable porta-potties. Most US cities have extremely limited access to public restrooms so the robo taxis get to serve double duty.
It will only take a small percentage of patrons to avail themselves of the opportunity for relief before potential customers are aware of the risk. And ultimately the customers who need waste disposal services will drive out the ones who don’t.
I don’t see any algorithm available to help the techies skip over this.
Widespread adoption of autonomous vehicles has always been a bizarre idea to me. It involves putting automation in the same cars and trucks that we’ve had for decades so they can drive across the same roads and highways that we’ve had for decades. There are only so many cars and trucks that you can put on these transportation arteries before traffic flows become unstable and break down into congestion or traffic jams. The point I’m getting at here is autonomous vehicles will not deliver anything faster, and it will not deliver more of anything over a given period of time – traffic will always restrict this.
What’s its really doing, particularly in an urban context, is replacing human drivers when there isn’t currently a shortage of them and the cost benefits are not substantial. Uber might want to replace its human drivers so it can collect the full fare for every trip, but then they will have to pay for all the transportation overhead that is currently covered by its drivers (vehicle lease payments, vehicle insurance, vehicle maintenance and repair, gasoline, road tolls, traffic tickets, and mobile phone bills). I’m not sure how much of a benefit they will get from implementing this technology – and that’s if it works.
Other transportation technologies, like railroads and steam ships and and the internal combustion engine, all provided massive boosts in transport efficiency for their time. Railroads replaced horse drawn carts because they could transfer more freight over longer distances over shorter periods – a lot more. Autonomous vehicles on current roads and highways, however, cannot increase the frequency of trips, or shorten delivery periods, or deliver over longer distances. They can only replace the driver in what is an inherently restricted network.
There might be a way to slightly increase this efficiency, but it is unclear if the cost associated with doing this would be prohibitively expensive. It would involve building autonomous vehicle only highways, and assigning a networking router like computer to manage them. The computer would control all the autonomous vehicles on this highway, simultaneously, and determine the most efficient way for every vehicle to collectively travel from their entry to exit ramp. It would function in the same way as computer routers that a manufactured by companies like Cisco, which determine the most efficient way for data packets to make they way through a telephone or computer network. How these vehicles get to the highway on ramp or to their final destination from the highway exit ramp might be managed by drivers at a remote location in the same way that unmanned aerial drones operate. The point here is safety and some efficiency – but in a way that may not be cost effective.
So while autonomous cars or trucks are a nice novelty and a curiosity, this technology might not be able to provide the same types of efficiencies as our transportation innovations from previous centuries.
Robo-taxis will only lead to Robo-bordellos.
Further to Kurtismayfield and TopHat, not only does cost recovery require high rider density – excluding suburbs and poorer urban areas – but most commentators also assume that drivers only drive. When are taxis busiest? – on party nights. Drivers keep vehicles clean, and potentially decline to give highly inebriated people into their vehicle to begin with. They also load/unload luggage, critical for airport runs, another area where taxis and their Uber/Lyft analogs do huge business. Ditto helping families and shoppers with stuff, not to mention infant seats. So Waymo and Cruise can save on driver costs, but then have higher cleaning costs at high-ridership times of the night when cleaning will take vehicles out of service, and there are sets of people who use taxis who won’t be able to use AVs.
Meanwhile drivers personalize their vehicles with all sorts of things, up to the Lincoln Lawyer extreme of Michael Connelly’s detective novels and subsequent movies. There are lots of use cases where Cruise/Waymo will not be a substitute for the personal(ized) vehicle.
The road to profitability will be long, when the focus is the taxi market. Long-haul trucking will be different, but again human drivers are hard to substitute away from for the first and last mile of a cross-continental run.
Not being F. Scott FitzGerald, I can’t hold two contradictory thoughts simultaneously and hope better minds than mine can help. On the one hand the CPUC has approved fully autonomous cars for robo-taxis. On the other hand, Tesla and Elon have admittedly that their fully autonomous mode is not ready and they are spending billions on AI and data centers to finish it. So what do Cruze and Waymo have that Tesla doesn’t? Is Tesla trying to get by with less hardware on the cars? I know there is a whistleblower lawsuit alleging Elon decreed a reduction of the LIDAR sensors on Teslas at the risk of worse autonomous performance. Is that what we’re looking at? I’m at a loss to reconcile the two assertions.
P
fully autonomous vehicles have a much more expensive suite of sensors afaik, so yes… and are designed from the ground up for full av, where teslas have always over-advertised assisted driving as autonomy.
I live near the beach in SF and this morning I had to go over to Geary/Masonic for an appointment. Construction was being done in the middle lane of Oak (along the Panhandle) and I got to see one of the Cruze/Waymo driverless vehicles come to a stop next to the construction and turn on the flashers and sit stationary. No one was playing tricks on it, it did it in an autonomous fashion. :-)
With that type of Skynet overlord, it seems like we ought to be able to successfully combat the AI takeover.