Forgot your password?
Close
typodupeerror
AI Transportation

Waymo Updates Vehicles to Better Handle Power Outages - But Still Faces Criticism (cnbc.com) 65

Posted by EditorDavid from the right-of-Waymo dept.
Waymo explained this week that its self-driving car technology is already "designed to handle dark traffic signals," and successfully handled over 7,000 last Saturday during San Francisco's long power outage, properly treating those intersections as four-way stops. But while during the long outage their cars sometimes experienced a "backlog" when waiting for confirmation checks (leading them to freeze in intersections), Waymo said Tuesday they're implementing "fleet-wide updates" to provide their self-driving cars "specific power outage context, allowing it to navigate more decisively."

Ironically, two days later Waymo paused their service again in San Francisco. But this time it was due to a warning from the National Weather Service about a powerful storm bringing the possibility of flash flooding and power outages, reports CNBC. They add that Waymo "didn't immediately respond to a request for comment, or say whether regulators required its service pause on Thursday given the flash flood warnings." And they also note Waymo still faces criticism over last Saturday's incident: The former CEO of San Francisco's Municipal Transit Authority, Jeffrey Tumlin, told CNBC that regulators and robotaxi companies can take valuable lessons away from the chaos that arose with Waymo vehicles during the PG&E power outages last week. "I think we need to be asking 'what is a reasonable number of [autonomous vehicles] to have on city streets, by time of day, by geography and weather?'" Tumlin said. He also suggested regulators may want to set up a staged system that will allow autonomous vehicle companies to rapidly scale their operations, provided they meet specific tests. One of those tests, he said, would be how quickly a company can get their autonomous vehicles safely out of the way of traffic if they encounter something that is confusing like a four-way intersection with no functioning traffic lights.

Cities and regulators should also seek more data from robotaxi companies about the planned or actual performance of their vehicles during expected emergencies such as blackouts, floods or earthquakes, Tumlin said.
This discussion has been archived. No new comments can be posted.

Waymo Updates Vehicles to Better Handle Power Outages - But Still Faces Criticism More

Waymo Updates Vehicles to Better Handle Power Outages - But Still Faces Criticism

Comments Filter:

  • There are 3 basic kinds of regulations:

    Safety, Ethical, and Busybody.

    Number of customers is NOT something that the government should be deciding. The reasonable number of autonomous vehicles on the city streets is exactly something that the market should decide, not government.

    Should it ramp up from a low number to something higher? Yes. But ramping up the number is something government does POORLY. Government works slowly because it is supposed to deal with the dangers and unethical issues, not busines

    • It absolutely 1000% is when you have something like autonomous vehicles which is something without precedent since we've had roads and motor vehicles. We all accept the government decides so much already of what happens on the roads via law, police, maintenance, design, why is this one thing off limits now?

      Government works as fast as it wants to operate or is allowed to operate. I think California has been rather quickly reactive to how they've handled self driving cars and part of why we can even have th

    • Number of customers is NOT something that the government should be deciding. The reasonable number of autonomous vehicles on the city streets is exactly something that the market should decide, not government... Government works slowly because it is supposed to deal with the dangers and unethical issues, not business issues.

      The view that things like "dangers" and "unethical issues" are separate and distinct form "business issues" is perhaps the most dangerous fallacy in all of Libertarianism.

      Business IS a danger, as is almost everything which is powerful and useful. Allowing powerful and useful things to be treated as morally and unquestionably superior and "right" - according them a hands-off regulatory approach - is a metastatic cancer in the body politic.

      Government damned-well should decide the "reasonable number of autono

    • There are 3 basic kinds of regulations: Safety, Ethical, and Busybody.

      A good percentage of regulations are made by grifters. The CEO of Simpson Strong-Tie was seen at NFL games in a box seat with the governor of California, and now suddenly all new houses need to have Simpson Strong-Tie hardware installed (not necessarily name brand, but they do have the patent). That sort of thing.

      • Re: (Score:2)

        by BranMan ( 29917 )

        Yeah, that's not grift. That is the CEO making use of his right to petition the government for redress of grievances. In this case that houses without hurricane (or typhoon) securing ties are not safe. Governor listened, agreed with the CEO, and decided to update the building codes on new construction accordingly.

        Happens all the time - no grift involved. California itself is not buying the ties, there are other manufacturers for them (they may have patents on their ties, but there are others that look d

        • It's grift lol.

          In this case that houses without hurricane (or typhoon) securing ties are not safe.

          Oh yeah? To protect against all the hurricanes in California? You are a tool.

  • It only handles the situations it's modeled for. Anything new fucks it over. By its very nature, it cannot handle anything new.

    • Re: (Score:2)

      by gweihir ( 88907 )

      That is also its strength. Applies to any form of automation: Once it is configured for something, it will handle a specific situation forever. Humans get tired, bored, arrogant, reckless, drunk, etc., automation does not.

      The tech in Waymo and other has a long way behind it. When I studied CS 35 years ago, I talked to somebody doing her Master's thesis on "2 lane left turn on an intersection with traffic lights" for self-driving cars. The tech was pretty impressive back then, but had obviously not much chan

      • Re: (Score:2)

        by narcc ( 412956 )

        FSD is fundamentally different from other kinds of automation.

        • Re: (Score:2)

          by gweihir ( 88907 )

          No. That is pure nonsense and FUD. It is in a higher difficulty-class, but that is it.

          • FSD is fundamentally different from other kinds of automation.

            No. That is pure nonsense and FUD. It is in a higher difficulty-class, but that is it.

            Well, if it isn't "fundamentally different from other kinds of automation" as narcc said, then isn't it at least two or three orders of magnitude more complex?

            Am I wrong in thinking that the numbers of combinations, permutations, and variables of real-world driving conditions are far, far higher than those in, say, factory automation?

            • Re: (Score:2)

              by gweihir ( 88907 )

              Factory automation can and has been done without computers. But no, real-world driving is not actually hard. I mean, most people can be qualified to learn it and do you know of anything else where that is possible?

              • Re: (Score:2)

                by narcc ( 412956 )

                real-world driving is not actually hard

                That's pretty delusional. If it's "not actually hard" then why don't we have FSD?

                • Re: (Score:2)

                  by gweihir ( 88907 )

                  You are just massively overestimating what evolution stage IT is at. Meshes nicely with your general cluelessness.

                  • Re: (Score:2)

                    by narcc ( 412956 )

                    You are just massively overestimating what evolution stage IT is at.

                    Umm... Aren't you the one here who is "overestimating what evolution stage IT is at"? I'm claiming that the technology is nowhere near as advanced as you're claiming.

                    Also, evolution does not work that way.

      • The tech is full of holes and you are looking at it as a tech maximilist with starry eyes.

        It's being alpha tested on public roadways and repeatedly failing. Without a solid legal liability framework around it.

        • Any proof it's "repeatedly failing"? It has been repeatedly successful. No at-fault deaths. Thousands of driverless taxis in densely populated San Francisco and zero at fault deaths or injuries and a per-mile accident rate far superior to humans. The one death that Waymo was involved in was when a human driven car hit a person and threw them against a Waymo .. so go figure!

          • Failures literally in this article. People have died. This is bigger than just Waymo.

            Way to narrow the problem until it's not a problem. Dimwits might fall for this style of argument, but it just makes you look foolish.

            • Re: (Score:2)

              by gweihir ( 88907 )

              Nice FUD you have there. Or rather lies. Because "people have died" not even a valid metric. Human drivers kill people every day. The only thing that matters is whether FSD kills more or less people. It kills significantly less people already.

            • Re: (Score:2)

              by dgatwood ( 11270 )

              Failures literally in this article. People have died. This is bigger than just Waymo.

              So you're judging all self-driving car tech by the worst? That's what regulations are for — to mandate standards that companies have to follow when developing their tech and shut their testing down when they don't do so.

          • Re: (Score:3)

            by gweihir ( 88907 )

            Exactly. FSD is still a bit limited and occasionally fails (and liars like Musk have been giving it a bad name), but it is already killing and maiming far fewer people per distance traveled than human drivers do. The problem with human drivers is that most are relatively bad while they think they are pretty good. And hence they reject FSD because it is worse than they _think_ they are, but it is already better in actual reality.

        • Re: (Score:2)

          by gweihir ( 88907 )

          No. I am looking at is as an actual IT expert that can see facts. You are just afraid of something you do not understand.

          This is NOT even Beta-testing. This is the optimization stage on the way to V2.0. Stop abusing terminology.

          • I do understand it. You are making excuses and have not even begun to refute my very simple and easy to understand post. For someone who fancies themselves an "expert" that should be deeply embarassing to you.

            • Re: (Score:2)

              by gweihir ( 88907 )

              You actually understand nothing. Nice. Dunning-Kruger left-side case, I see.

              So, to be clear. No, the tech is not "full of holes". That is a direct lie. It needs to be very much NOT "full of holes" to even be allowed on the streets. Also note that the tech has been independently reviewed in quite a few more places. Second, please read up what "alpha test" actually means. Third, the legal framework in place is on the same level as human drivers, or rather better because you can actually sue a company with som

    • It has some reasoning capability that's why it's called AI and not just programming of rules. The thing is Waymo uses very little AI compared to Tesla.

    • To be fair, people struggle in this area too. "My travel agent job was automated in the 1990s, what do I do for a living now?" There are a million jobs that have been automated, through mechanization, through computerization, and now through AI. People who used to do those jobs have trouble figuring out how to find or do new jobs.

  • Serious problem (Score:5, Insightful)

    by abulafia ( 7826 ) on Saturday December 27, 2025 @04:09PM (#65884979)
    I live in SF. The power outage was probably a blessing in disguise, causing this highly visible failure when it was "just" a power outage.

    Waymo failed spectacularly [fox9.com] at handling this.

    Real emergencies not uncommonly also cause power failures. So at the exact the time you really need, say, firetrucks to get around or get people out of somewhere, all these fucking robots become bollards.

    They really need a true fail-safe mode tested to actually work for these things, pay to have the human capacity on the backend to handle emergency conditions, or be limited in numbers such that they can't shut down a city. Maybe this needs federal oversight, there are some obvious national security implications to a private facility that can remotely shut down mobility in major metros.

    And it can always get worse [wikipedia.org], if you've read that novel...

    • Re: (Score:2)

      by gweihir ( 88907 )

      So what? This tech is in its last optimization phase. In a few years it will be able to handle expected but rare situations as well and be generally better than an average human driver.

      • Re: (Score:2)

        by narcc ( 412956 )

        This tech is in its last optimization phase

        Last? That seems ... optimistic. Uncharacteristically optimistic, if I'm not mistaken.

        In a few years it will be able to

        It's always "in a few years". Elmo has been making similar predictions for 10 years now. I'm still waiting for the magic car that will pay for itself working for Uber when I'm not using it.

        • Re: (Score:2)

          by gweihir ( 88907 )

          Thank you for making it clear you have neither arguments nor insight into the question.

          • Re: (Score:2)

            by narcc ( 412956 )

            Sorry? All you did was make a bold assertion. If you want an argument, you're going to need to do better than that!

            What makes you think that "This tech is in its last optimization phase"? Or do you "have neither arguments nor insight into the question"?

      • So what? This tech is in its last optimization phase.

        Probably not. It puts on a good show, much like chatbot neural networks do.

    • Spectacularly?

      Footage posted by TikTok user @brahmsstan and obtained by Storyful shows several Waymo vehicles stalled at an intersection in the Mission District neighborhood.

      While the Waymo Driver is designed to treat non-functional signals as four-way stops, the sheer scale of the outage led to instances where vehicles remained stationary longer than usual to confirm the state of the affected intersections. This contributed to traffic friction during the height of the congestion.

      I dunno man. Even a s

      • 100% agree. People have been fucking that situation up for ages.

      • Re: (Score:2)

        by abulafia ( 7826 )
        There was a lot of chaos, and it took Waymo six hours to clear it all.

        I personally saw it at 5th and Market - 3 Waymos were blocking all three northbound lanes. Human drivers can let a firetruck through until the road is hopeless. The robots just block them.

        Maybe mandatory human overrides on these things. If the theft risk it too great, well, maybe they aren't ready for prime time.

        • Re: (Score:2)

          by Jeremi ( 14640 )

          Human drivers can let a firetruck through until the road is hopeless. The robots just block them.

          Robots can also let a firetruck through until the road is hopeless -- they just need to be programmed to do so. Assuming Waymo isn't completely insensate, that will be the case going forward.

          • Re: (Score:2)

            by flink ( 18449 )

            Robots can also let a firetruck through until the road is hopeless -- they just need to be programmed to do so. Assuming Waymo isn't completely insensate, that will be the case going forward.

            Humans can adopt to a situation on the fly and exercise common sense. Waymo goes into a failsafe mode and becomes a roadblock, potentially getting people killed. It can only adopt after the fact. After the damage has been done. Telling the family of the guy who dies on the way to the hospital because a bunch of Waymos are blocking the road that "you'll fix it next time" is cold comfort.

            • Re: (Score:2)

              by Jeremi ( 14640 )

              Waymo goes into a failsafe mode and becomes a roadblock,

              Not if Waymo is clever (which they are). A Waymo's car can and should go into a failsafe mode where it maneuvers its way out of traffic and parks, and also moves out of the way of emergency vehicles when possible/necessary. Basically the same things a responsible human being would do in the same situation. It's not rocket science.

          • Human drivers can let a firetruck through until the road is hopeless. The robots just block them.

            Robots can also let a firetruck through until the road is hopeless -- they just need to be programmed to do so. Assuming Waymo isn't completely insensate, that will be the case going forward.

            I see no evidence it isn't already the case. There was no footage of emergency vehicles being blocked.

        • Human drivers can let a firetruck through until the road is hopeless. The robots just block them.

          How do you know they wouldn't have moved for a fire truck? Waymos use external microphones to listen for and react to emergency vehicles, as well as cameras to watch for them. Unless you somehow know that their programming would not have prioritized moving out of the way of emergency vehicles over waiting for confirmation of how to treat an intersection with a light out, you're just assuming problems where you don't actually know they exist.

      • The issue seems to be the car got confused and just stopped in the middle of the road. It should have pulled over to the side and waited for normal operations to resume. That should be fairly easy to fix assuming it can handle a four way stop sign.situation.

    • Re: (Score:2)

      by Cyberax ( 705495 )
      Eh. Waymo will be fixed and it will be programmed to "safe" itself in case a widespread event occurs. Probably by driving until it finds a place to pull over and wait for the event to resolve.

      If you look at development of ANY new large-scale technology (trains, cars, planes), it's always the same. You'll have hilarious early failures, catastrophes, unforeseen confluences of events leading to disasters, etc. But eventually all the glaring bugs get fixed.
      • This is happening on public streets with no real legal liability framework around it. Stop looking at it as a starry eyed tech maximilist and start looking at it in terms of real people living and working in these areas.

        • Re: (Score:2)

          by Cyberax ( 705495 )

          This is happening on public streets with no real legal liability framework around it.

          Oh, puhlease. So far, Waymos have not hurt a single person. And the worst outcome was a bit less gridlock than during any regular blackout.

          All the news outlets are basically bloviating about movie scenarios like: "But what if it's a Godzilla invasion and we need to make sure Voltron vehicles can quickly pass through".

          • So far, Waymos have not hurt a single person.

            Liar. (I assume you're not too stupid to use a search engine, so the probably explanation is you are actively lying here).

            • Re: (Score:2)

              by Cyberax ( 705495 )
              Go on, give me the list of people hurt by Waymo. I'll wait. Still waiting. Waiting. OK, let me google it for you: https://www.reddit.com/r/SelfD... [reddit.com]

              So far, Waymo indeed hasn't hurt anybody.
              • From your link:

                December, 2024, Atlanta, GA On December [XXX], 2024 at 5:48 PM ET a Waymo Autonomous Vehicle ("Waymo AV") operating in Atlanta, Georgia made contact with the pavement at the elevated entrance to [XXX]. near [XXX]. The Waymo AV was traveling south on [XXX] when it approached [XXX] ahead on its left. As the Waymo AV turned left onto [XXX], the Waymo AV's undercarriage made contact with the raised pavement of the side walk where [XXX] meets [XXX]. At the time of the contact, the Waymo AV's Level 4 ADS was engaged in autonomous mode, and a test driver was present (in the driver's seating position). The test driver in the Waymo AV reported a moderate injury.

                • Re: (Score:2)

                  by Cyberax ( 705495 )
                  No airbag deployment. "Moderate injury" can be anything. So yeah, Waymo so far hasn't hurt anybody. I guess some mechanic also might have hurt themselves while servicing it?
    • Yes, this outage sure was a blessing as it showed another situation which wasn't properly handled, but that's the beauty of robocars, you just fix it with one update over the complete fleet of cars and use the incident as another startingpoint to think of other situations you might have not thought about and fix those before they happen and create cars that are much safer as human drivers (in general). Yes, there will then still be situations which robocars won't handle properly, but those can be fixed one
      • How many people using public streets or depding on them for emergency services to arrive are an acceptable level of loss for this technology to be beta tested on public streets? How many dollars in property damage?
        • You mean like how many human drivers every day keep emergency vehicles from arriving on time? These robocars will handle an emergency vehicle better as most human drivers will.

    • Real emergencies not uncommonly also cause power failures.

      Real emergencies also cause complete breakdown of any taxi service, Waymo or otherwise, as well as most private transport too. Nothing learned about this will apply to "real emergencies". You're not going to be calling your Waymo during the apocalypse.

      • What is this kind of "rhetoric" called, where you intentionally miss the point and make up a wildly over the top example that never happens while ignoring all the other scnerios that happen often and absolutely apply to contradict the point you're trying to make?

  • They now call them 'driverless' cars instead of autonomous cars or self-driving cars, but yet again we see that a lot of the time they are merely remote-controlled cars.

    And out of the woodwork will come the shills shouting about 'self-driving cars are already safer than human drivers', while avoiding the question of "how can they be better than human drivers when they hand over to human drivers for the tricky bits"?

    They are still in development, how safe will they be when they are commoditised and are handi

    • They now call them 'driverless' cars instead of autonomous cars or self-driving cars, but yet again we see that a lot of the time they are merely remote-controlled cars.

      I would like more transparency on how often humans do take control of Waymos.

  • Yes, AI will take many jobs. But the process will be very slow, and will come in fits and starts. AI companies, like Waymo, will have to figure out how to engineer their way around new obstacles they didn't anticipate in their original design. Waymo won't be wiping out Uber driver jobs on a massive scale, overnight. Robotaxis will take a long, long time to be able to fully replace human drivers, just as it took a long time for the automobile to replace the carriage driver. Time is a good thing, it gives hum

  • When the going gets tough, Waymo stops going.

Slashdot Top Deals

"The trouble with doing something right the first time is that nobody appreciates how difficult it was." -- Walt West

Close