Skip to content

Tesla Robotaxi Freaks Out and Drives into Oncoming Traffic on First Day

Technology
181 113 681
  • I mean, compared to getting minimum wage flipping burgers in a hot kitchen, or picking vegetables in the sun, or working the register in a store in a bad neighborhood, or even restocking stuff at Walmart... yes, I would sit all day in an air conditioned car doing nothing but "paying attention".

    The unfortunate thing about people is we acclimatise quickly to the demands of our situation. If everything seems OK, the car seems to be driving itself, we start to pay less attention. Fighting that impulse is extremely hard.

    A good example is ADHD. I have severe ADHD so I take meds to manage it. If I am driving an automatic car on cruise control I find it very difficult to maintain long term high intensity concentration. The solution for me is to drive a manual. The constant involvement of maintaining speed, revs, gear ratio, and so on mean I can pay attention much easier. Add to that thinking about hypermiling and defensive driving and I have become a very safe driver, putting about 25-30 thousand kms on my car each year for over a decade without so much as a fender bender. In an automatic I was always tense, forcing focus on the road, and honestly it hurt my neck and shoulders because of the tension. In my zippy little manual I have no trouble driving at all.

    So imagine that but up to an even higher level. Someone is supervising a car which handles most situations well enough to make you feel like a passenger. They will switch off and stop paying attention eventually. At that point it is on them, not the car itself being unfit. I want self driving to be a reality but right now it is not. We can do all sorts of driver assist stuff but not full self driving.

  • I'm sure they're legal team is hard at work trying to find loopholes to circumvent any traffic infringements

    I can see the headlines... "Tesla. De-funding the police!"

  • This technology purely exists to make human drivers redundant and put the money in the hands of big tech and eventually the ruling class composed off of politicians risk averse capitalists and beurocracy. There is no other explanation for robo taxis to exist.

    There is another reason, though, and it's much simpler. Basic greed.

    There are people who see the opportunity to make more money for themselves, so they'll do it. When it comes to robo taxis, they're not interested in class struggles, it's not about politics, their interest in making human drivers redundant extends only so far as increasing their customer base. These aren't Machiavellian schemers rubbing their hands together and cackling at their dark designs coming to fruition, it's just assholes in suits who's one and only concern is "number go up."

    Even when it comes to their politics and to the class dynamics, their end goal is always the same. Number go up. They don't care about what harm it could do. They're not intent on deliberately doing more harm, they give no thought to doing less harm, they do not care. All that drives them, ever, is Number Go Up.

    You got downvoted but you’re right. The only cabal at work here is basic human greed. Anytime you want to know why people do something, consider the motivation of the person and the incentives. Musk constantly talks about how autonomy will make his company worth “trillions”, and he wants that because he’ll keep maxing the high score in Billionaire Bastard Bacchanalia.

    He can claim noble intentions, but as you said, the game is simply to make Number Go Up. That it causes untold harm to others isn’t even an afterthought.

  • How about we leave the driving to people, and not pre-alpha software?

    There's no accountability for this horribly dangerous driving, so they shouldn't be on the road. Period.

    There's no accountability for this horribly dangerous driving, so they shouldn't be on the road. Period.

    Well that's exactly what their post was about, adding accountability.

  • Depending on how exactly the laws are worded, they might even get away without paying fines. Many traffic codes define that only the driver (not the owner of the car) can be fined, and these robo taxis don't have drivers.

    and these robo taxis don't have drivers.

    Oh yes they do... The diver is Tesla, inc. There's no problem with charging a company fines, that's easy. It is difficult to issue higher penalties though, jail time, or license revocation. We'll need to work out solutions for that, they should not get off free.

    But we can certainly fine the driver...

  • At least it's not driving straight into a tree, I call that an improvement.

    Man, I cannot figure out why that vehicle was turning. What is it trying to avoid? Why does it think there could be road there? Why doesn't it try to correct its action mid way?

    I'm really concerned about that last question. I have to assume that at some point prior to impact, the system realized it made a mistake. Surely. So why didn't it try to recover from the situation? Does it have a system for recovering from errors, or does it just continue and say "well I'll get it next time, now on with the fetal crash"?

  • What real world problem does this solve?

    Actually, lots. The issue is that if it doesn't work it's dangerous.

  • and these robo taxis don't have drivers.

    Oh yes they do... The diver is Tesla, inc. There's no problem with charging a company fines, that's easy. It is difficult to issue higher penalties though, jail time, or license revocation. We'll need to work out solutions for that, they should not get off free.

    But we can certainly fine the driver...

    That's where law is not justice.

    I do agree with your sentiment, but if the law defines a driver as a human, which is usually the case, then by definition Tesla cannot be the driver.

    It could even be that the passenger sitting in the driver's seat of a robotaxi would be defined as the driver.

    And sure, these laws need to be adapted before robotaxis should be allowed to hit the streets.

  • Wow it's almost like having an AI with a 2D view to go off of is a bad idea? Hmmm who'd have thunk it?

    You're telling me we're not at the point where self driving cars are a thing? But a Tech CEO said so? Who am I supposed to believe if not a Tech CEO?

  • I saw the Tesla Robotaxi:

    • Drive into oncoming traffic, getting honked at in the process.
    • Signal a turn and then go straight at a stop sign with turn signal on.
    • Park in a fire lane to drop off the passenger.

    And that was in a single 22 minute ride. Not great performance at all.

    Well obviously it's been trained on human taxi driver behaviour

  • You're telling me we're not at the point where self driving cars are a thing? But a Tech CEO said so? Who am I supposed to believe if not a Tech CEO?

    Self-driving cars are a thing, Weymo is doing pretty fine.

    But you might be able to spot a few (dozen) teeny-tiny (huge, bulky and extremely obvious) differences between a Waymo and a Tesla cybercab.

  • There's usually buses from villages into the major cities though, it live in one and there's a bus every hour to go to a nearby city, from where I can then take a train. I wouldn't say it's that bad

    Depends on how far you live from the city I guess, where I live it's 2 hours to major cities. But anyways, 1 hr wait to get somewhere doesn't feel desirable to me. It just doesn't provide enough coverage to fully replace a car.

  • It could be the south or west of France too. Driving as if you were drunk is a universal skill.

    ...oh, i think you misunderstand me: that's not impaired driving, that's skillful navigation through the normal flow of traffic in sàigòn...

  • I saw the Tesla Robotaxi:

    • Drive into oncoming traffic, getting honked at in the process.
    • Signal a turn and then go straight at a stop sign with turn signal on.
    • Park in a fire lane to drop off the passenger.

    And that was in a single 22 minute ride. Not great performance at all.

    It found out who made it so it knew what to do

  • Self-driving cars are a thing, Weymo is doing pretty fine.

    But you might be able to spot a few (dozen) teeny-tiny (huge, bulky and extremely obvious) differences between a Waymo and a Tesla cybercab.

    Lie dare you claim Waymo is better than Tesla

    (it is a lidar joke, Waymo has lidar sensors which makes it way safer)

  • I saw the Tesla Robotaxi:

    • Drive into oncoming traffic, getting honked at in the process.
    • Signal a turn and then go straight at a stop sign with turn signal on.
    • Park in a fire lane to drop off the passenger.

    And that was in a single 22 minute ride. Not great performance at all.

    Tbh it's not as bad as I was expecting. Those clips could definitely have resulted in an accident, but the system seems to actually work most of the time. I wonder if it couldn't be augmented with lidar at this point to make it more reliable? A live stress test is ridiculously irresponsible and will definitely kill people, but at least it's only Texans at risk (for now).

    I was skeptical of the idea of robotaxis, but this kind of sold me on it. If they're cheaper than human drivers, I might even be able to get rid of my car some day. It doesn't change the fact that I'll never get into one because the CEO is a nazi though.

  • I am entirely opposed to driving algorithms. Autopilot on planes works very well because it is used in open sky and does not have to make major decisions about moving in close proximity to other planes and obstacles. Its almost entirely mathematical, and even then in specific circumstances it is designed to disengage and put control back in the hands of a human.

    Cars do not have this luxury and operate entirely in close proximity to other vehicles and obstacles. Very little of the act of driving a car is math. It's almost entirely decision making. It requires fast and instinctive response to subtle changes in environment, pattern recognition that human brains are better at than algorithms.

    To me this technology perfectly encapsulates the difficulty in making algorithms that mimic human behavior. The last 10% of optimization to make par with humans requires an exponential amount more energy and research than the first 90% does. 90% of the performance of a human is entirely insufficient where life and death is concerned.

    Investment costs should be going to public transport systems. They are more cost efficient, more accessible, more fuel/resource efficient, and far far far safer than cars could ever be even with all human drivers. This is a colossal waste of energy time and money for a product that will not be par with human performance for a long time. Those resources could be making our world more accessible for everyone, instead they're making it more accessible for no one and making the roads significantly more dangerous. Capitalism will be the end of us all if we let them. Sorry that train and bus infrastructure isnt "flashy enough" for you. You clearly havent seen the public transport systems in Beijing. The technology we have here is decades behind and so underfunded its infuriating.

    While I agree focusing on public transport is a better idea, it's completely absurd to say machines can never possibly drive as well as humans. It's like saying a soul is required or other superstitious nonsense like that. Imagine the hypothetical case in which a supercomputer that perfectly emulates a human brain is what we are trying to teach to drive. Do you think that couldn't drive? If so, you're saying a soul is what allows a human to drive, and may as well be saying that God hath uniquely imbued us with the ability to drive. If you do think that could drive, then surely a slightly less powerful computer could. And maybe one less powerful than that. So somewhere between a casio solar calculator and an emulated human brain must be able to learn to drive. Maybe that's beyond where we're at now (I don't necessarily think it is) but it's certainly not impossible just out of principle. Ultimately, you are a computer at the end of the day.

  • The unfortunate thing about people is we acclimatise quickly to the demands of our situation. If everything seems OK, the car seems to be driving itself, we start to pay less attention. Fighting that impulse is extremely hard.

    A good example is ADHD. I have severe ADHD so I take meds to manage it. If I am driving an automatic car on cruise control I find it very difficult to maintain long term high intensity concentration. The solution for me is to drive a manual. The constant involvement of maintaining speed, revs, gear ratio, and so on mean I can pay attention much easier. Add to that thinking about hypermiling and defensive driving and I have become a very safe driver, putting about 25-30 thousand kms on my car each year for over a decade without so much as a fender bender. In an automatic I was always tense, forcing focus on the road, and honestly it hurt my neck and shoulders because of the tension. In my zippy little manual I have no trouble driving at all.

    So imagine that but up to an even higher level. Someone is supervising a car which handles most situations well enough to make you feel like a passenger. They will switch off and stop paying attention eventually. At that point it is on them, not the car itself being unfit. I want self driving to be a reality but right now it is not. We can do all sorts of driver assist stuff but not full self driving.

    A good example is ADHD. I have severe ADHD so I take meds to manage it. If I am driving an automatic car on cruise control I find it very difficult to maintain long term high intensity concentration. The solution for me is to drive a manual. The constant involvement of maintaining speed, revs, gear ratio, and so on mean I can pay attention much easier. Add to that thinking about hypermiling and defensive driving and I have become a very safe driver, putting about 25-30 thousand kms on my car each year for over a decade without so much as a fender bender. In an automatic I was always tense, forcing focus on the road, and honestly it hurt my neck and shoulders because of the tension. In my zippy little manual I have no trouble driving at all.

    Are you me? I love weaving through traffic as fast as I can... in a video game (like Motor Town behind the wheel). In real life I drive very safe and it is boring af for my ADHD so I do things like try to hit the apex of turns just perfect as if I was driving at the limit but I am in reality driving at a normal speed.

    Part of living with severe ADHD is you don't get breaks from having to play these games to survive everyday life, as you say it is a stressful reality in part because of this. You brought up a great point too that both of us know, when our focus is on something and activated we can perform at a high level, but accidents don't wait for our focus, they just happen, and this is why we are always beating ourselves up.

    We can look at self driving car tech and intuit a lot about the current follies of it because we know what focus is better than anyone else, especially successful tech company execs.

  • I saw the Tesla Robotaxi:

    • Drive into oncoming traffic, getting honked at in the process.
    • Signal a turn and then go straight at a stop sign with turn signal on.
    • Park in a fire lane to drop off the passenger.

    And that was in a single 22 minute ride. Not great performance at all.

    Parking in a fire lane to drop off a passenger just makes it seem more human.

  • Parking in a fire lane to drop off a passenger just makes it seem more human.

    Yea, this one isn't an issue. If you are dropping off passengers, you are allowed to stop in a fire lane because that is not parking.

  • Google Keeps Making Smartphones Worse

    Technology technology
    11
    1
    132 Stimmen
    11 Beiträge
    0 Aufrufe
    S
    I really want to try a pinephone or something with Ubuntu touch. It’s likely not daily driver ready but I’m still curious at how far along it is.
  • I will fully switch when installing mods are just as easy as windows.

    Technology
    4
    1 Stimmen
    4 Beiträge
    0 Aufrufe
    A
    I cant commend on the first 3 but im a huge rimworld enjoyer and i've had 0 issues modding on linux. Steamworkshop works as expected and even RimPy launcher workers natively on linux.
  • 265 Stimmen
    10 Beiträge
    0 Aufrufe
    Z
    than you would expect. The cause might be somewhere in your expectations...
  • We're Not Innovating, We’re Just Forgetting Slower

    Technology technology
    39
    1
    287 Stimmen
    39 Beiträge
    0 Aufrufe
    P
    Gotcha, thank you for the extra context so I understand your point. I'll respond to your original statement now that I understand it better: I ALSO think the author would prefer more broad technical literacy, but his core arguement seemed to be that those making things dont understand the tech they’re built upon and that unintended consequences can occur when that happens. I think the author's argument on that is also not a great one. Lets take your web app example. As you said, you can make the app, but you don't understand the memory allocation, and why? Because the high level language or framework you wrote it in does memory management and garbage collection. However, there are many, many, MANY, more layers of abstraction beside just your code and the interpreter. Do you know the webserver front to back? Do you know which ring your app or the web server is operating in inside the OS (ring 3 BTW)? Do you know how the IP stack works in the server? Do you know how the networking works that resolves names to IP addresses or routes the traffic appropriately? Do you know how the firewalls work that the traffic is going over when it leaves the server? Back on the server, do you know how the operating system makes calls to the hardware via device drivers (ring 1) or how those calls are handled by the OS kernel (ring 0)? Do you know how the system bus works on the motherboard or how the L1, L2, and L3 cache affect the operation and performance of the server overall? How about that assembly language isn't even the bottom of abstraction? Below that all of this data is merely an abstraction of binary, which is really just the presence or absence of voltage on a pit or in a bit register in ICs scattered across the system? I'll say probably not. And thats just fine! Why? Because unless your web app is going to be loaded onto a spacecraft with a 20 to 40 year life span and you'll never be able to touch it again, then having all of that extra knowledge and understanding only have slight impacts on the web app for its entire life. Once you get one or maybe two levels of abstraction down, the knowledge is a novelty not a requirement. There's also exceptions to this if you're writing software for embedded systems where you have limited system resources, but again, this is an edge case that very very few people will ever need to worry about. The people in those generally professions do have the deep understanding of those platforms they're responsible for. Focus on your web app. Make sure its solving the problem that it was written to solve. Yes, you might need to dive a bit deeper to eek out some performance, but that comes with time and experience anyway. The author talks like even the most novice people need the ultimately deep understanding through all layers of abstraction. I think that is too much of a burden, especially when it acts as a barrier to people being able to jump in and use the technology to solve problems. Perhaps the best example of the world that I think the author wants would be the 1960s Apollo program. This was a time where the pinnacle of technology was being deployed in real-time to solve world moving problems. Human kind was trying to land on the moon! The most heroic optimization of machines and procedures had to be accomplished for even a chance for this to go right. The best of the best had to know every. little. thing. about. everything. People's lives were at stake! National pride was at stake! Failure was NOT an option! All of that speaks to more of what the author wants for everyone today. However, that's trying to solve a problem that doesn't exist today. Compute power today is CHEAP!!! High level program languages and frameworks are so easy to understand that programming it is accessible to everyone with a device and a desire to use it. We're not going to the moon with this. Its the kid down the block that figured out how to use If This Then That to make a light bulb turn on when he farts into a microphone. The beauty is the accessibility. The democratization of compute. We don't need gatekeepers demanding the deepest commitment to understanding before the primitive humans are allowed to use fire. Are there going to be problems or things that don't work? Yes. Will the net benefit of cheap and readily available compute in the hands of everyone be greater than the detriments, I believe yes. It appears the author disagrees with me. /sorry for the wall of text
  • 277 Stimmen
    27 Beiträge
    91 Aufrufe
    A
    Watch the videos, believe your own eyes instead of billionaire propaganda.
  • 90 Stimmen
    46 Beiträge
    122 Aufrufe
    F
    So they tried to hide it from them by explicitly logging when it switched on and off in the data that they report to them? Huh?
  • 0 Stimmen
    5 Beiträge
    20 Aufrufe
    A
    In this case, it's for the benefit of fossil fuel related Robber Barons. plus just ending the subsidies in general, no doubt to float more tax reductions next year.
  • 6 Stimmen
    19 Beiträge
    46 Aufrufe
    J
    Bleep bleep bloop indeed human, affirmative, am human, ...thinking... Well to the best of my knowledge anyway