A thread to discuss self-driving cars and other vehicles. No politics, please.
Technology, commercial aspects, legal considerations and marketing are all on-topic.
- Companies (e.g. Tesla Inc.) are only on-topic when discussing their self-driving products and research, not their wider activities. The exception is when those wider activities directly impact (or are impacted by) their other business areas - e.g. if self-driving car development is cut back due to losses in another part of the business.
- Technology that's not directly related to self-driving vehicles is off-topic unless you're discussing how it might be used for them in future.
- If we're talking about individuals here, that should only be because they've said or done something directly relevant to the topic. Specifically, posts about Tesla do not automatically need to mention Elon Musk. And Musk's views, politics and personal life are firmly off-topic unless you can somehow show that they're relevant to self-driving vehicles.
Cadillac plans to release a scaled back, more simple version of similar technology by 2015 - what they call "Super Cruise", which isn't total self-driving, but does let you relax on highways. It positions your car in the exact center of a lane, slows down or speeds up as necessary, and is said to be meant for ideal driving conditions (I'm guessing that means ideal weather, no rain or snow, etc.).
I am looking forward to such tech. If enough people prefer to drive this way, and the technology works reliably, it could result in safer roads with fewer accidents. Another possibility is that, using GPS and maybe the ability to know ahead of time which roads are most clogged, they can find the quickest route from place to place.
On the other hand, hacking could be a real concern, and I hope it doesn't become a serious threat. It's looking like we're living more and more like those sci-fi Everything Is Online worlds depicted in fiction for a long time.
(Mod edited to replace original post)
Edited by Mrph1 on Mar 29th 2024 at 4:19:56 PM
Figure of Hourai
Presumably, you ought to do one whilst doing the other.
Avatar Source
Maybe start by not pitching the system in such a way that makes people think it's OK to sleep at the wheel.
"Yup. That tasted purple."
Lost in Space
Ah, we're back here. Figured someone would post that. As is my trademark, I'm going to list out actual facts to inform this... opinionated discussion.
- NHTSA conducts lots of investigations that do or do not result in recalls. It's curious that the media is choosing to broadcast this particular one, but there is a lot of public interest in Tesla and in self-driving, so we can give it a pass.
- There is, curiously, no information given in the reporting that would compare rates of crashes with first responders between Tesla Autopilot and other Level 2/3 ADAS software. I feel like these would be germane to its relative safety.
- There is also no reporting on the relative rates of crashes with first responder vehicles that occur while cars are being driven manually as a point of comparison. In other words, are you more or less likely to do that while driving manually or while on Autopilot?
- The report notes that at least half of the crashes occurred in conditions that Tesla says Autopilot is not reliable in, meaning drivers are using it out of spec.
- The "disengagements less than one second" prior to the crash thing is a red herring. When investigating and reporting Autopilot crashes, Tesla counts any event that occurs within five seconds of a disengagement as happening under Autopilot. It is not passing off responsibility here, but the reporting fails to note this.
Most importantly, though, the version(s) of Autopilot that NHTSA is investigating are fairly old. If this were any other automaker, the model year and other factors would be relevant, but every Tesla on the road capable of running Autopilot can and has already been updated. A "recall" is nonsensical; it'd be like Apple issuing a recall for iOS 8 or something. The latest version is 15.5! And yes, Tesla has already updated Autopilot to add various safety features to reduce or eliminate first-responder crashes.
Driver attentiveness while using ADAS is a big deal, obviously, and there have been tests of the effectiveness of monitoring systems across a wide variety of such systems. Some are better than Autopilot, some are worse. Tesla is fully capable of having its system be more aggressive: FSD Beta uses its driver-facing camera with AI face detection to gauge attentiveness and it is reportedly very strict, but that is beta software and not suitable for deployment to every vehicle.
In the end it comes down to a few questions:
- Is Autopilot safe when used in an intended manner?
- Has Tesla adequately addressed the causes of these crashes through currently deployed software updates?
- Is Tesla marketing Autopilot in such a way as to create the impression in a reasonable person that it can be used without paying attention?
- Is Tesla adequately preventing such abuse?
These matters will be determined by NHTSA, not in the court of public opinion.
Edited by Fighteer on Jun 12th 2022 at 10:28:14 AM
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"
What does a reasonable person think “autopilot” means then?
"Yup. That tasted purple."
> These matters will be determined by NHTSA, not in the court of public opinion.
In terms of actual requirements on Tesla, sure.
But people, such as here, can and will form their own opinions about the matter based on what evidence is known, and may even make buying/investing decisions based on it. That doesn't require a full NHTSA report and there's nothing wrong about that.
Waiting to have *any* kind of opinion isn't even a scientific position. That would be, 'this is what's known today and I'm aware of what's currently known, so this is my best informed opinion on what's going on. If what's known changes, so might my opinion on the matter.' You can't wait until all the evidence is in because all the evidence is *never* in, there is always more to know.
Visiting from the Hoag Galaxy
![[up]](https://static.tvtropes.org/pmwiki/pub/smiles/arrow_up.png "[up]"){kind=icon}
Yeah, that's the big question. What does the average person think autopilot means, how good is Tesla's documentation about it and how clear is Tesla about the limitations of their self-driving software?
Is there a message that pops up whenever you turn it on saying it must never be left unattended? It's still in beta, so encouraging people to turn it on and stop paying attention isn't a good idea, if only because glitches and errors can happen. How easy is it for someone who's really excited and who thinks Tesla can do no wrong to gloss right over any warnings or safety measures?
On top of that, there should be an investigation into that thing where it shuts off immediately before an impact. It's really bad optics that it does that on a frequent enough basis to get noticed, for one, but also, if it is a safety feature kicking in, it's a really poorly designed one that's not activating when it should. It should err way on the side of caution in terms of handing control back over, not at the literal last second. And the optics do matter. Because whether or not its intentional, it looks like it shuts off immediately before impact so the impact can't be registered as being due to the use of autopilot.
Yes, this. At this point, I will never buy a Tesla, because a lot of red flags keep popping up. If it was just a few, I could maybe ignore them, but at this point, it really looks like the company has really wonky priorities and standards in a bunch of ways and I would rather get an EV from a company that doesn't pointlessly fuck with safety features, that can figure out how to do a windshield wiper on a truck and that doesn't keep trying to reinvent the wheel on things where doing that has no benefit or point.
Edited by Zendervai on Jun 12th 2022 at 11:14:31 AM
Not Three Laws compliant.
Lost in Space
This is a "gotcha" question, because it's not just about the name but the marketing as a whole. On Tesla's website, it is advertised as a driver assistance system and its capabilities are clearly stated. When you buy a Tesla, you are provided with instructions on its features, including the proper use of Autopilot. When you engage it, you are shown warnings about proper use. If you fail to maintain attention, it warns you and then turns off.
So the question is: would a reasonable person driving a Tesla, in the face of all this information, come to the belief that Autopilot allows unsupervised self-driving? Could a person who intentionally defeats its driver monitoring features have a reasonable claim to be unaware of the need to supervise it?
Backing off from Tesla, do ADAS features in general lead to drivers having a false sense of security and paying less attention than they otherwise would?
Remember that there are different things being discussed here. NHTSA is investigating Autopilot: Tesla's standard level 2 ADAS that provides adaptive cruise control and lane-keeping on highways. There are also Enhanced Autopilot (sold in some markets with a limited set of FSD features), the regular Full Self-Driving package (that you get access to when you buy the feature), and the FSD Beta package.
that thing where it shuts off immediately before an impact.
If the driver suddenly becomes alert and takes control of the vehicle, that is a disengagement, and that is what the report is talking about. This is what I mean about people having poorly informed opinions — and about the media inaccurately describing the facts.
Edited by Fighteer on Jun 12th 2022 at 11:21:07 AM
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"
Given the number of times people have been caught sleeping at the wheel, I’d say that they’re believing that Autopilot allows totally unsupervised self-driving.
"Yup. That tasted purple."
Lost in Space
But whose fault is that? This is where regulators and courts need to step in.
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"
If Tesla is plainly explaining the features of the product and what it can and cannot do, why are people still making assumptions about what it can do?
"Yup. That tasted purple."
Visiting from the Hoag Galaxy
x4 Like I said, it's an optics thing. It looks really bad and saying "no, the driver took control at the last second" sounds like an excuse. Because it's super easy to flip the meaning of that sentence.
The driver took control at the last second before the impact can mean one of two things. It can either mean the driver initiated it, or it could mean the computer initiated it. Both scenarios result in the driver taking control at the last second. This doesn't mean that the computer is designed to give up control in case of a collision, but it does mean that it needs to be properly investigated.
A ton of people are dumb as hell and don't listen to a thing anyone says, preferring to follow the narrative in their own heads. There are people who try to break into stores and restaurants after hours because they ignored the closed signs and all the evidence, that kind of person is going to assume that it's a true autopilot.
That being said, it might be necessary for Tesla to put it behind a barrier of some sort, to prevent people from just mindlessly switching it on. If it requires active intent, that'll block at least some of the people who just ignore everything that doesn't fit their own internal script.
Edited by Zendervai on Jun 12th 2022 at 11:22:25 AM
Not Three Laws compliant.
Because there is a certain fraction of human beings who will always be idiots. The question is what does the average Tesla driver do? ![[nja]](https://static.tvtropes.org/pmwiki/pub/images/user_ninja_2074.png "[nja]"){kind=small}
Edited by DeMarquis on Jun 12th 2022 at 11:22:59 AM
I think there’s a global conspiracy to see who can get the most clicks on the worst lies
Lost in Space
But are people doing that, or are these outliers? Again, that's what NHTSA needs to figure out. I have never spoken with a Tesla owner (in person or online) who believes that Autopilot lets the car drive unsupervised, but my sample is obviously not statistically valid or representative.
This is what I mean about the responsibility of the media to report accurately. I don't blame people for being misinformed if they are being given bad information by people who should know better.
Also, Tesla does put a disclaimer up on the in-car display when you activate Autopilot for the first time, and as we've discussed, it will disable the feature if it detects misuse from the driver.
Edited by Fighteer on Jun 12th 2022 at 11:24:36 AM
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"
Visiting from the Hoag Galaxy
A disclaimer showing up once isn't good enough. Do you really think anyone will really pay attention to that? No one reads the license agreement or the EULA on anything, why would you expect people to pay attention to a screen that shows up one time, possibly while driving that's presumably just a "tap yes if you understand this" type thing?
And yes, it is on Tesla to make it as clear as possible how it's supposed to work and what the risks are. People improperly using the feature is incredibly dangerous, especially since it's still being developed, and "I'm an incredibly oblivious person" is a defense that's really hard to make an argument against in many cases.
Not Three Laws compliant.
Lost in Space
In the current legal and regulatory environment, the driver is responsible for the car at all times when it is using a Level 2 ADAS. We might not be doing enough to educate people about that.
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"
Figure of Hourai
This discussion hasn't materially changed since the last time it came up. I still hold that calling it Autopilot is an incredibly stupid move on Tesla's part if your interest is in safety or clear communication and not marketing.
Avatar Source
Visiting from the Hoag Galaxy
I second that. If it's only supposed to be driver assistance, they should have just called it that. Autopilot has a really specific meaning to most people and deliberately banking on people having a positive association with that meaning is backfiring on Tesla right now.
Branding has an enormous impact on how people perceive things, and choosing a name that doesn't actually match the product is a real problem.
Edited by Zendervai on Jun 12th 2022 at 11:56:46 AM
Not Three Laws compliant.
Lost in Space
A German court did hold that the Autopilot brand is deceptive. As of yet, no US agency or court has ruled on the matter. The pro line is that everyone understands that autopilots do not fly planes without human supervision.
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"
Visiting from the Hoag Galaxy
Except this isn't actually true. A lot of people think a plane only needs a pilot for emergencies and that the plane could fly just fine without one. Because no one talks about it, but there are stories of planes being able to land themselves in an emergency. It's not actually a reasonable assumption to make, that everyone understands how it actually works.
And coming to the conclusion that planes basically fly themselves and that Tesla's developing a similar system is very easy, specifically because Tesla named it the same thing as the thing that many people assume is already an autonomous control system.
Edited by Zendervai on Jun 12th 2022 at 12:41:01 PM
Not Three Laws compliant.
Lost in Space
But are there real people who believe this or is it just a hypothetical used as a strawman? As I said, I have not met any actual person who does not understand the distinction. Granted, my personal experience is not a representative sample of the general public.
Edited by Fighteer on Jun 12th 2022 at 12:42:03 PM
"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"
The people who are disregarding Tesla instructions are definitely a thing and not a strawman, they are after all the issue here!
"For a successful technology, reality must take precedence over public relations, for Nature cannot be fooled." - Richard Feynman
Visiting from the Hoag Galaxy
Um...yes. There are real people who think that.
Just like there are real people who think that customer service people are literal robots. And I have met people who think that. When it comes to Tesla, Fighteer, your circle is seemingly mostly people who have a vested interest in them.
How many people do you regularly interact with who have no knowledge of electric vehicles or any of the mechanics behind cars or transit or anything like that? How many of them do you regularly talk to about stuff like this?
There are people who genuinely think the Earth is fucking flat or that the moon is fake or that chemtrails are dispensing mind control drugs. Why is the idea of underinformed people assuming that a thing called autopilot is an actual autopilot a step too far?
Never assume the average person is too smart to hold a given opinion, you will always be disappointed. You can always find a disturbingly large number of people who hold the most insane, idiotic, dangerous opinions if you bother to look.
Edited by Zendervai on Jun 12th 2022 at 12:47:10 PM
Not Three Laws compliant.
Lost in Space
Okay, and are companies responsible for protecting people like that from every possible consequence of their ignorance?
Figure of Hourai
When it's because they've co-opted a term that already existed in the public consciousness? Yes.
Avatar Source
The agency document all but says Tesla’s method of making sure drivers pay attention isn’t good enough and that it has a safety defect that should be recalled, said Bryant Walker Smith, a University of South Carolina law professor who studies automated vehicles.
A better suggestion: Replace it with a system that can solve the problem even if the driver is asleep.
"For a successful technology, reality must take precedence over public relations, for Nature cannot be fooled." - Richard Feynman