Please create an account to participate in the Slashdot moderation system

 


Forgot your password?
Close
typodupeerror
×
AI Transportation

After Firetruck Crash, California Tells Cruise to Reduce Robotaxi Fleet by 50% in San Francisco (sfchronicle.com) 160

Posted by EditorDavid from the taxi-driverless dept.
Thursday a Cruise robotaxi drove through a green light in front of an oncoming firetruck "with its forward facing red lights and siren on, the San Francisco Police Department said in a statement to Reuters." The San Francisco Chronicle adds that the Cruise vehicle's passenger "passenger was treated on the scene and shared taken in an ambulance to a hospital, though the company said the injuries were 'non-severe.' The company added in an email to the Chronicle that the passenger was on the scene walking around and talking to emergency responders before being taken to the hospital."

By Friday California's Department of Motor Vehicles said it was investigating the "concerning incidents," according to TechCrunch. But it adds that the AV-regulating agency also "called for Cruise to reduce its fleet by 50% and have no more than 50 driverless vehicles in operation during the day and 150 driverless vehicles in operation at night until the investigation is complete. Cruise told TechCrunch it is complying with the request. Cruise also issued a blog post giving the company's perspective of how and why the crash occurred.
Cruise's blog post points out the firetruck was unexpectedly in the oncoming lane of traffic that night. But meanwhile, elsewhere in the city... The same night, a Cruise car collided with another vehicle at 26th and Mission streets. The company said another driverless car, which had no passengers, entered the intersection on a green light when another car ran a red light at high speed. The driverless car detected the other car and braked, according to Cruise, but the two cars still collided...

The collisions came a day after city officials asked state regulators to halt their approval of robotaxi companies' unrestricted commercial expansion in the city, citing concerns about how the robotaxis' behavior impacts emergency responders.
Last weekend Cruise was also criticized after "as many as 10 Cruise driverless taxis blocked two narrow streets," reports the Los Angeles Times: Human-driven cars sat stuck behind and in between the robotaxis, which might as well have been boulders: no one knew how to move them.... The cars sat motionless with parking lights flashing for 15 minutes, then woke up and moved on, witnesses said.
Cruise "blamed cellphone carriers for the problem," according to the article — arguing that a music festival overloaded the cellphone network they used to communicate with their vehicles.

Thanks to Slashdot reader jjslash for sharing the story.
This discussion has been archived. No new comments can be posted.

After Firetruck Crash, California Tells Cruise to Reduce Robotaxi Fleet by 50% in San Francisco More

After Firetruck Crash, California Tells Cruise to Reduce Robotaxi Fleet by 50% in San Francisco

Comments Filter:

  • passenger needs to sue and cellphone network???? (Score:5, Insightful)

    by Joe_Dragon ( 2206452 ) on Saturday August 19, 2023 @09:50AM (#63779988)

    passenger needs to sue
    but cellphone network is needed for the cars and network issues can lead to crashes?

    an driver less car needs to work with NO NETWORK as you have
    dead zones
    roaming areas with low speed caps
    areas with low max speeds
    networks with slow down after XX data used
    networks with high lag.

    • Re:passenger needs to sue and cellphone network??? (Score:5, Funny)

      by test321 ( 8891681 ) on Saturday August 19, 2023 @10:02AM (#63780026)

      In The Phatom Menace (Lucas 1999) and Oblivion (Kosinski 2013) droids fall to the ground after they lose connection to the mother ship and I thought what a stupid movie nobody would ever design autonomous robots like this. Then some years after robotaxis crash when they lose mobile phone connectivity.

    • Cellphone network is likely just needed for remote control, for when one of infinite corner cases which make urban driving AI-hard comes up. These things need to be nudged in the right direction occasionally to get them out of a bind.

      • Clearly they haven't tested the "network not available" scenario thoroughly enough. That should be the default testing mode.

        • When remote control is needed and there is no network it's stuck, no amount of testing will help. It's the inevitable result of partial autonomy.

          • Re: (Score:3)

            by danda ( 11343 )

            and partial autonomy is all that will ever be achieved. These things are not sentient. At some point VC money is going to run out and society is going to decide this driverless car thing was worth trying but ultimately doomed to failure.

    • Yes, I can't believe they are trying to blame the network carriers. If your car can't safely operate without a network connection it should pull over and not operate when it detects a network failure. Better yet it should not have an permit to operate without the ability to safely drive with no connection.

  • Lies and propaganda (Score:5, Insightful)

    by phantomfive ( 622387 ) on Saturday August 19, 2023 @09:50AM (#63779990) Journal

    Cruise told TechCrunch it is complying with the request. Cruise also issued a blog post giving the company's perspective of how and why the crash occurred.

    The blog post doesn't explain why the crash occurred. It is a blog post that says, "We did everything right and it was not our fault somehow it happened."

    Cruise "blamed cellphone carriers for the problem," according to the article — arguing that a music festival overloaded the cellphone network they used to communicate with their vehicles.

    Amazing. Everything bad that happens is not their fault. Who could have guessed that cell phone connectivity might fail?

    • I was thinking the same thing. Damn this company just has the worst luck. *rolls eyes*

    • > Who could have guessed that cell phone
      > connectivity might fail?

      Well, in first world countries basic infrastructure and utilities like telephone service are meant to be able to be relied upon and are NOT supposed to fail except in the most extreme of natural disasters. And even then, the expectation is that utilities are amongst the first thing to be fixed and reactivated. Cruise may have screwed up, sure. But, if a disinclination on their part to provide the advertised level of service for which

      • Re:Lies and propaganda (Score:4, Insightful)

        by haruchai ( 17472 ) on Saturday August 19, 2023 @11:12AM (#63780190)

        If the cars can't operate without reliable connectivity they are NOT autonomous, they're remote-controlled

      • Re:Lies and propaganda (Score:5, Insightful)

        by Registered Coward v2 ( 447531 ) on Saturday August 19, 2023 @11:14AM (#63780194)

        > Who could have guessed that cell phone > connectivity might fail?

        Well, in first world countries basic infrastructure and utilities like telephone service are meant to be able to be relied upon and are NOT supposed to fail except in the most extreme of natural disasters. And even then, the expectation is that utilities are amongst the first thing to be fixed and reactivated. Cruise may have screwed up, sure. But, if a disinclination on their part to provide the advertised level of service for which Cruise and the public pay for contributed to the issue, that doesn't mean the telcos should be excused from bearing the blame for their own failures to properly provision and maintain their own service and infrastructure.

        If you design your system to rely on 100% cell phone network availability and then turn it loose on city streets it's not the telco's fault. Congestion and deprioritization are well known facets of the infrastructure's operation, if Cruise wanted 100% priority then they should do what companies that want uninterruptible power an pay extra for it. Even then, power outages happen, cables get cut, Taylor Swift comes to town and a system that relies on the network should be designed to account for cell phone outages. It sounds to me like Cruise just wants to shift the blame, and instead pointed out a flaw in their design.

        • > Taylor Swift comes to town

          I actually went to the Eras tour, in two different cities actually. It was fantastic. It was epic. You know what it wasn't? Unplanned. She doesn't just show up in town on a whim, announce a flash mob on social media, and expect locals to cope on zero notice. I bought my tickets nine months in advance of the concerts. That means the shows were planned, with the venues secured, all contracts signed, and capacities known, long before ticketbastard's own inexcusable fuckup

          • Just like the Eras Tour, San Francisco is a known quantity. Outside Lands is a known quantity. For any infrastructure provider to be unprepared for any or the three is, again, inexcusable. Safety on the streets of San Francisco I can tell you is no longer a known quantity, and thus firetruck runs are most often responses to medical conditions of the homeless. That is why everyone who can afford to move away, is moving away. And no all intersections are not Stroads (4+1center) crossing with Stroads (4+1

          • Outside Lands is a known quantity. For any infrastructure provider to be unprepared for any or the three is, again, inexcusable. Yes, Cruise does deserve criticism for their failure. But don't pretend like they're alone.

            You don't build out infrastructure for a worse case peak demand scenario that happens infrequently, you manage the traffic with some getting slowed or have higher latency. While Cruise may not be alone, they bear the major share of the blame if their software couldn't handle known network behavior.

      • Well, in first world countries basic infrastructure and utilities like telephone service are meant to be able to be relied upon and are NOT supposed to fail except in the most extreme of natural disasters.

        False. Cell phone networks fail all the time.

      • Well, in first world countries basic infrastructure and utilities like telephone service are meant to be able to be relied upon and are NOT supposed to fail except in the most extreme of natural disasters.

        I take it you've never been to a football game or a concert or out of your mother's basement in a first world country. No the first world doesn't magically solve an inherent problem with LTE, and cellular services fail constantly, even in countries that regularly top rankings for best infrastructure.

      • in first world countries basic infrastructure and utilities like telephone service are meant to be able to be relied upon and are NOT supposed to fail except in the most extreme of natural disasters.

        In capitalist countries basic infrastructure and utilities like telephone service are meant first and foremost to make money, and are NOT penalized for failing to be available during even for the most trivial of reasons.

        If you don't have a contract with a telco that guarantees uptime, guess what? Yeah, you know where this is going.

        In reality, failing to take facts into account when doing engineering work leads to failures.

    • Oddly enough, I'll take "the cell phone networks were down, throwing the self driving car into a dumber local-only mode, etc..." as an explanation for the crash, much like "The driver was so drunk he passed out and drove off the road into a tree" is an explanation. They aren't excuses though.

      The difference? An explanation is a reason, an excuse is a waiver of blame/liability.

      If they're smart, the response to this will eventually be "we have added this scenario to our training sets and are boosting process

  • A strange decision... (Score:5, Insightful)

    by david.emery ( 127135 ) on Saturday August 19, 2023 @09:50AM (#63779994)

    So if an autonomous vehicle has a problem that causes it to crash into emergency vehicles, it seems to me that reducing the number of those does NOT address the underlying fault. So sure, statistically, the chance of an accident is cut in half with a 50% reduction in autonomous vehicles. That's small consolation for anyone in the back of the remaining 50% of vehicles that continue to have this problem..

    • Re: (Score:2)

      by gweihir ( 88907 )

      Indeed. Was my first thought also. What the hell is reducing number of flawed vehicles going to do about the flaw?

      • Reduces the chances of it happening again at least until the investigation is complete and/or the company has submitted that it has fixed the problem so it won't happen again.

        Banning them - 100% reduction, would obviously virtually eliminate the odds of it happening again, but the city might have decided that that is a step too far.

        In addition, fewer cars means that at least theoretically the company can put more attention towards each car. Plus, if they were a traditional transport company, perhaps operat

        • 50% makes no sense as safety policy. This would be like allowing half the drunk people in a bar to drive themselves home.

    • The fire truck crashed into the Cruise. Fire trucks crash into cars all the time, for example: https://www.carscoops.com/2022... [carscoops.com]

  • With driverless cars, the standard has to be that they are better than a human driver, not that they are never involved in an accident. A fire engine may be allowed to run red lights but surely it has a duty to give way to other traffic that may not have noticed it? Every time I see an emergency vehicle running a red light they always slow down and go through cautiously for exactly that reason.

    Similarly being hit by some idiot running a red light happens to human drivers too and, as the article mentioned

    • Re: (Score:2)

      by at10u8 ( 179705 )
      The passenger is alive, so I will wager that the fire truck did slow down.

      • The passenger is alive, so I will wager that the fire truck did slow down.

        In my experience they do at intersections to avoid accidents like the one that happened.

    • Re: (Score:2)

      by jeremyp ( 130771 )

      In the UK, the fire engine would be liable in this situation. They are doing something that is inherently dangerous so they have the duty of care.

      • In the UK, the fire engine would be liable in this situation.

        It's really hard to say that since unfortunately we don't know what happened. I tried figuring out what intersection it happened at but couldn't find that either.

  • Computer driven cars are still in their infancy and are already safer than humans albeit slightly more annoyingly. Human drivers have peaked, traffic accidents by human driver murder 40,000 Americans every year. One million humans worldwide die EVERY year, because a human driver is either temporarily distracted or stupid.

    The only hope of reducing traffic fatalities is for computers to drive the vehicles of anyone but the most elite drivers. I understand a lot of sicko people don't care about the one million

    • Re: (Score:2)

      by rossdee ( 243626 )

      " If US deaths can be reduced from 40,0000 a year to 39,999 a year it would still be worth it."

      Reducing traffic deaths by a factor of 10 would certainly be worth it.

      • We would all drive M1 tank chassis speed limited to 4mph if this were true. Because 4 people are killed per year in my state by trees actively falling into roadways does not mean we need to cut down all trees within 80 feet of a roadway. If it takes 40,000 man hours to save 1 human life it is break even to society. 2000 hours (1 Work Year) x 20 years. Some poor engineer just wasted his lifetimes work to save an idiot who could not be bothered to look both ways, or save the family who chose the jo

    • Computer driven cars are still in their infancy and are already safer than humans albeit slightly more annoyingly.

      [Citation needed]. A good citation, not one that comes from company propaganda.

      If US deaths can be reduced from 40,0000 a year to 39,999 a year it would still be worth it.

      You don't actually believe this, you are lying. If you really wanted to reduce traffic deaths, you would be advocating for breathalyzers to be installed on all vehicles for operation, to prevent drunk driving.

      • I'm good with that, but you and I both a large majority of America will never agree to that.

        • Since when has you ever moderated your opinion based on what a large majority of America thinks? This is new.

          • Wasting time convincing people and government to force breathalyzers on every driver would take away energy from telling people to allow self driving cars. Also, self driving cars will save a lot more lives than forced breathalyzers would. Eliminating drunk driving will reduce traffic deaths by nearly half in the best case, but enabling full self driving can reduce deaths by 90% or more.

      • If you really wanted to reduce traffic deaths, you would be advocating for breathalyzers to be installed on all vehicles for operation, to prevent drunk driving.

        If vehicle interlock breathalyzers didn't have false positives, that would be a reasonable stance. Guess what?

        • Re: (Score:2)

          by haruchai ( 17472 )

          "If vehicle interlock breathalyzers didn't have false positives,"
          what is the rate of false positives?

    • Computer driven cars are the only path to reducing traffic accidents.

      Untrue. While I will agree that it is a good path to reducing traffic accidents, it's hardly the "only" path. There's proper engineering of roadways, for example, in that there are roads where we can estimate that somebody will die there every year. Other roads don't have this.
      Identify the problems with the road, fix them, and you're better off.

      For example, in one of my old towns there was a underpass with a measurable fatality rate. They replaced the intersections with roundabouts(traffic circl

  • Yes the fact that they rely on cell coverage to operate is already very bad but the concept that if the car reaches a type of fault state where it stops moving and doesn't know enough to find a place to park safely is egregious.

  • Good driving is expecting the unexpected (Score:5, Insightful)

    by Nkwe ( 604125 ) on Saturday August 19, 2023 @10:03AM (#63780032)

    Cruise's blog post points out the firetruck was unexpectedly in the oncoming lane of traffic that night.

    Why should this be unexpected? If it is unexpected, it is a serious oversight. It is standard procedure for emergency vehicles (with the emergency lights and siren on) to go around vehicles stopped at a red light by using the oncoming traffic lane.

    Aside from the fact that emergency vehicles with flashing lights and sirens blaring should be expected to drive where most drivers don't and such behavior should be expected, unexpected things do happen when driving. The concept of "defensive driving" is all about expecting the unexpected and as a driver, paying attention to what other drivers are doing and expecting that other drivers will make mistakes.

    Do Cruise vehicles not have audio input to listen for sirens and have their visual models trained to look for emergency vehicles lights? If not, it is a serious oversight and they probably shouldn't be on the road.

    • Do Cruise vehicles not have audio input to listen for sirens and have their visual models trained to look for emergency vehicles lights?

      Yes. They claimed to have recognized the emergency vehicle but still crashed into it.

    • This, they identified the emergency vehicle "almost immediately", detected the siren, and then broke the law by not coming to a stop. Instead they continued to drive until recognizing an accident was impending, then still didn't stop but slowed.

      It doesn't matter where an emergency vehicle is, or what it is doing, for safety and compliance with law, you come to a controlled stop, clearing the path if you can, so they can do anything they need/want to.

      This isn't the time for "oh they are in x lane so I can",

      • It doesn't matter where an emergency vehicle is, or what it is doing, for safety and compliance with law, you come to a controlled stop, clearing the path if you can

        Especially true if you can hear the vehicle but can't see where it is.

      • Re: (Score:2)

        by g01d4 ( 888748 )

        It doesn't matter where an emergency vehicle is, or what it is doing, for safety and compliance with law, you come to a controlled stop, clearing the path if you can

        "If you can" can be subjective. Especially in these types of situations, and it's not like humans never get into accidents with emergency vehicles. And it's not like the person driving the emergency vehicle always drives in the best fashion as some might think sirens and lights confer magical powers.

        I think it does matter where the vehicle is.

        • Your responsibility is to clear the roadway, and stop until emergency traffic has passed (or until emergency responders direct you to move). Do it safely, but do it immediately.

          You are not expected to predict that someone else will respond poorly. Everyone is responsible for their own actions. Yes, someone else failing to respond properly may harm you. The fact that that is possible does not authorize you to respond inappropriately "just in case". Trying to beat the firetruck (or ambulance, or police c

    • It also seems like one of the easiest safety protocols to program. Sensors should have no issue detecting those distinctive bright lights and the car should stop.
    • Well hindsight is easy, but during development it could be something they just didn't think of, like many 'unexpected situations'. But, at least it should, once a situation like that occured, one would need to add it to the capabilities of the driverless system, and ofcourse immediately try to think other situations like it. I even think creators of driverless systems should have access to a realtime accidentreport system so real life situations, caused by other driverless systems/human drivers, can be anal

  • ... when immature tech is beta tested in public
    While I have confidence that it will eventually be perfected, it's a really, really, REALLY hard problem
    And yes, I have worked on self-driving software for a major manufacturer

    • While I have confidence that it will eventually be perfected

      Why do you have that confidence? What is it based on?

  • The details are a bit sparse but the explanation from Cruise [getcruise.com] has some clues:

    The AV positively identified the emergency vehicle almost immediately as it came into view, which is consistent with our underlying safety design and expectation. It is worth noting, however, that the confines of this specific intersection make visual identification more challenging – for humans and AVs alike – as it is significantly occluded by buildings, meaning that it is not possible to see objects around the corner

    • If I'm designing an AV my default "oh-shit" response has the AV stopping, of course if you're in an intersection you're now at risk of getting t-boned. This is a predictable problem, but the solution "drive to safety" could easily turn into "AV inexplicably swerves onto sidewalk and mows down small child".

      You have to "Drive to safety" because the expected behavior when an emergency vehicle is trying to come through is "pull out of the way and stop". If the self-driving vehicles can't handle that at least as safely as a human driver, then once again, they simply don't belong on the road yet.

      • If I'm designing an AV my default "oh-shit" response has the AV stopping, of course if you're in an intersection you're now at risk of getting t-boned. This is a predictable problem, but the solution "drive to safety" could easily turn into "AV inexplicably swerves onto sidewalk and mows down small child".

        You have to "Drive to safety" because the expected behavior when an emergency vehicle is trying to come through is "pull out of the way and stop".

        True, but I think they do have that behaviour down.

        The failure here wasn't misbehaving around an emergency vehicle, it was not detecting the emergency vehicle. When it finally detected the emergency vehicle (thing about to hit it) in the middle of the intersection it went into collision avoidance mode and stopped.

        If the self-driving vehicles can't handle that at least as safely as a human driver, then once again, they simply don't belong on the road yet.

        I think these are more dangerous for passengers because I suspect they haven't nailed the safe default for an intersection.

        I think they're safer for pedestrians since they always pay attention and

      • What if the AI driver is just better than 51% of human drivers at 5:40pm after a daylight saving time change? Long term attention to detail in respective actions where past success happened by random chance is not a human strong point. We are not there yet on self driving outside of specific conditions. The problem is the group of 51% of human drivers drifts minute by minute much more than we want to admit to ourselves. I say that with my only accident in the past million miles is a deer, and traini

    • Cruise AVs have the ability to detect emergency sirens, which increase their ability to operate safely around emergency vehicles and accompanying scenes. In this instance, the AV identified the siren as soon as it was distinguishable from the background noise.

      And yet it did nothing. Did it not think the siren was loud enough? Does it not track the delta to realize it's getting louder?

      It sounds like its sound detection process is flawed. I usually hear a siren at a far enough distance that I have plenty of time to stay out of its way; and my hearing is not all that great. A siren is an unusual noise in the background and humans react to background noises that don't normally belong; it was how we learned to survive around potential predators. I wonder if the AV simply treated the low siren as background noise until it was so loud it was obvious what it was; a better approach would be t

      • They really didn't describe the situation very well at all.

      • Humans are the ones that are restricted from listening in a 360-degree amplified pattern.

        Why would we NOT outfit an autonomous car with an external array of microphones to greatly enhance the machines ability to listen well beyond a human limitation?

        We'd be showing our human ignorance otherwise.

      • Cruise AVs have the ability to detect emergency sirens, which increase their ability to operate safely around emergency vehicles and accompanying scenes. In this instance, the AV identified the siren as soon as it was distinguishable from the background noise.

        And yet it did nothing. Did it not think the siren was loud enough? Does it not track the delta to realize it's getting louder?

        It sounds like its sound detection process is flawed. I usually hear a siren at a far enough distance that I have plenty of time to stay out of its way; and my hearing is not all that great. A siren is an unusual noise in the background and humans react to background noises that don't normally belong; it was how we learned to survive around potential predators. I wonder if the AV simply treated the low siren as background noise until it was so loud it was obvious what it was; a better approach would be to recognize sirens in background noise and act accordingly. It seemed to leave out the "what is that low sound I hear, sounds like a siren, where is it coming from" analysis.

        I suspect it heard the siren, the problem the AV has is deciding what it should do in response to the siren.

        The majority of sirens we hear involve emergency vehicles that never cross our paths. We'll drive a bit more cautiously and if it gets loud enough we'll slow down and even pull over, even if we don't see the vehicle because we know if it's that loud it's really close. An AV that pulled over every time it heard a siren would be perpetually parked.

        For audio volume it's often less "how loud is X" and "ho

    • To be clear, I like the idea they're allowing testing, but they need to realize it's not a "see if the vehicles are ready" test. It's "they're just barely safe enough to allow testing to improve the tech" testing. And for that I'm not sure they need 50/150 vehicles running at once for that.

      Figure that these vehicles are like 1% as smart as humans, learn at 1% of the speed*. Because they're relatively dumb, you need much closer scenarios in the training data in order to correctly identify and respond to the various edge cases. IE humans are able to generalize better.

      In any case, identifying all the edge cases can take a LOT of time. Ergo, more cars to get it done in parallel.

      Then figure that you have a half dozen or so self driving initiatives all operating in SF, and it can get messy at 10

      • To be clear, I like the idea they're allowing testing, but they need to realize it's not a "see if the vehicles are ready" test. It's "they're just barely safe enough to allow testing to improve the tech" testing. And for that I'm not sure they need 50/150 vehicles running at once for that.

        Figure that these vehicles are like 1% as smart as humans, learn at 1% of the speed*. Because they're relatively dumb, you need much closer scenarios in the training data in order to correctly identify and respond to the various edge cases. IE humans are able to generalize better.

        In any case, identifying all the edge cases can take a LOT of time. Ergo, more cars to get it done in parallel.

        True, but how is much of the training is data for the ML and how much is "hey, Alive and Eve, work together and fix intersection behaviour".

        If it's ML data then sure you need a lot. But if it's work for the devs then they've already got more than they can handle.

        The other thing is they should maybe be training these things in small cities, not SF.

        Driving in SF is constant sirens, j-walkers, weird lane situations, etc, etc.

        Test them in small cities instead. You've got low traffic, few sirens, few music festi

    • To be clear, I like the idea they're allowing testing, but they need to realize it's not a "see if the vehicles are ready" test. It's "they're just barely safe enough to allow testing to improve the tech" testing. And for that I'm not sure they need 50/150 vehicles running at once for that.

      If they're in that kind of testing mode, it's fine, but they need a safety driver behind the wheel, ready to take control at all times.

  • If Cruise can't move its POS, I can. If Cruise's passengers don't like that they can jump out.

    These things still have human controls, right? Is something stopping a human from getting in the driver's seat and acting like a driver?

  • A lot of drivers out there are much worse.

  • Cruise "blamed cellphone carriers for the problem," according to the article — arguing that a music festival overloaded the cellphone network they used to communicate with their vehicles.

    Teslas don't use the cellphone network when in autonomous driving mode, yet they keep plowing into parked emergency vehicles with lights on [thestreet.com].

  • Why are we rushing to approve AI driverless car technology? It's not ready for real world, large scale deployment. When you can't trust AI to write a professional and respectful email, why are we allowing it to drive with no guidance around a city?

    You can make the argument that people get into accidents and cause traffic issues, but at least then you can hold something and someone accountable, and if required revoke the license. When an AI decides to "park the car", how do you hold someone accountable?
  • Was the effect of LiDAR exposure on skin and eyes thoroughly tested?

    • Well known to turn people into super mutants.

      YMMV, of course.

    • Was the effect of LiDAR exposure on skin and eyes thoroughly tested

      The effects on humans are well known becasue we get doses of it every day from the Sun. Not to mention the satellites using it to image the ground.

      However, since you asked, this tidbit [velodynelidar.com] has a few more details.

  • If Cruise cars only work if they can communicate with the mother ship, are they truly autonomous? Shouldn't they just be called remotely controlled vehicles?

  • Most of the bad press around AVs seems to involve Cruise. Waymo has occasional problems as well, but it seems to be for minor things like getting caught in dense fog, or getting stuck due to unexpected road closures and crowds of pedestrians. Maybe Cruise should take a pause and let Waymo continue to deploy?

  • Do cruiseicles have ears to hear a siren? And recognize it as such?

    • Re: (Score:2)

      by Jahta ( 1141213 )

      Yeah, that was my first thought too. When driving I always hear emergency vehicles before I see them, and when I hear the siren I'm immediately extra vigilant.

  • It isn't autonomous is it needs a cellular connection.

    Ive worked in cellular for 20 years .. the KPIs around connectivity (time active on LTE/NR) are above 99% in most areas of the US .. but never 100%.

    e.g. these autonomous cars _will_ hit moments of no connectivity.

    • I had to do signal mapping of a rather large region once to determine which carrier was best for the regional government to go with.

      I can tell you this - the 'coverage maps' the mobile providers publish are not anywhere near granular enough and there are plenty of dead spots if you look. And I found no one carrier was able to sufficiently cover the region so we chose the best and made note of the dead zones.

      And even in the good areas, towers can be flooded by traffic... and by the way, emergency services o

  • Cruise "blamed cellphone carriers for the problem," according to the article — arguing that a music festival overloaded the cellphone network they used to communicate with their vehicles.

    This answer by Cruise is utter bullshit. No "autonomous" car should need to phone home to resolve a potential problem. If that were the case, the cars are not autonomous because a mobile phone connection, never mind the Internet connection that piggy backs on it, are never guaranteed. Besides, does the automation i

  • The burden of perfecting a LUXURY SALES FEATURE should fall on the manufacturers not the public.

    Self-driving is far from a social necessity, demonstrated by well over a century of human-controlled vehicles. It's a toy a bauble to amuse those too lazy to drive and too self-centered to care about their quality participation in the traffic stream.

    Unfortunately the public are far too stupid to demand useful regulation so some will die and be maimed during beta testing.

  • Why? (Score:2)

    by PPH ( 736903 )

    Cruise "blamed cellphone carriers for the problem,"

    Does Cruise need to receive some real time data about emergency vehicle operations over the cellular network? And if so, is this an example of the cellular companies sabotaging peer-to-peer V2X [wikipedia.org] technology with their own cellular based crap?

Slashdot Top Deals

Aren't you glad you're not getting all the government you pay for now?

Close