I feel that the hype for self-driving cars is the same as it was for VR in the 90s

We had to wait another 15-20 years for those feverished VR dreams to become a reality

I reckon it'll be the same for self-driving. We are 20-25 years away, not 5-10

In addition, only a few days (one or two weeks) ago we had a much commented story linked here on HN that VR is way too hyped. Supprted by quotes from actual executives and/or investors in that business if I recall correctly, so not just the usual negativism that can always be found for any topic if you choose your source well.

I don't think VR will ever be mainstream unless hardware costs come down drastically and they can figure out a way to make VR headsets look less dorky

In 5-10 years i see self-driving on freeways and maybe other main roads that have standard junctions, markings and signs. Maybe with lanes only for self-driving.

A self driving vehicle would do the long-haul and a human would do the tricky city/country lane leg of the journey.

Automating the whole journey is a long way off as you say.

And even if we do get self-driving cars, people forget that there will be an FDA clinic trial like process before they are allowed on the roads, which will take years.

Multiply that by the number of states/countries where you want to operate and by the level of bureaucracy in each country.

I'd say that from the day the tech actually works properly (which might be many years away, still), it will take at least 10 years for its availability across the globe, if not more.

Yeah, just like how rigorously Tesla's auto pilot has been vetted?

How long do you think they're going to be allowed to call it "Autopilot" when it's clearly not, and that confusion causes customers to trust it in ways that they shouldn't, and end up dying?

I'm not saying Tesla should be stopped or that Elon should go to jail. But I think it's a clear case of false advertising, where lives are on the line.

Actual airplane "autopilots" can be something as simple as automatic wing leveling or altitude hold. Using that name for Tesla's technology is only inaccurate in that they don't make airplanes. Where did everybody get the idea that "autopilot" implies full autonomy?

I expect an inflatable driver to appear and take the wheel, requiring only an occasional top up of air to keep things on the level.

I actually wonder if Airplane! is where people got the idea that "autopilot" means "the plane entirely flies itself." The bit where it's an inflatable doll is obviously satire, but maybe people thought the capabilities were real and the satire was just the doll part?

Sure, let's torture language for Tesla's benefit! Why not? I mean, we shouldn't let a silly thing like people's lives get in the way.

Since you know about historical autopilots I suspect this will be redundant, but here it is anyhow: https://en.wikipedia.org/wiki/Autopilot#Modern_autopilots

"Modern" autopilots have been around for a while, something like 40-50 years. Sure if you go back 80 years you can find autopilots that aren't as advanced. If you want to move the goalposts like that, relative to what a reasonable person might expect, knock yourself out.

What part of that article do you think is a point against what I'm saying, exactly? Autoland and taxiing? That requires a Category III airport which many airports, even those served by airliners, aren't. It also requires pilot supervision and pilot intervention in the event of emergency. Hardly autonomous!

And there are plenty of systems below that. For example:

http://sarasotaavionics.com/category/autopilots/certified-au...

Note: single axis and 2-axis. Just like what I said it can be.

But yes, the people insisting that "autopilot" does not imply autonomy, simply because no autonomous autopilot exists on any commercial aircraft, are torturing the language. Sure.

Find ~10 normal people and ask them what an autopilot does on an airplane. They're all going to say something along the lines of "it flies the airplane" and their intuition lines up with reality for about 90%+ of flights that are done with autopilots.

Ask those same ~10 people what an autopilot does for a car. They'll say something like "it drives the car". Given what the Tesla Autopilot (tm) is capable of doing, their understanding of what the name means and what the software can actually do only has maybe a 20% overlap.

So yes I would call that torturing language. It's an autopilot that can drive the car under a very limited set of circumstances AND it requires that Jesus (you, the driver) is always 100% alert and ready to snatch the wheel the second the car starts to do an unsafe thing.

It's a totally different control problem in cars than it is in airplanes, and it's impressive that they're doing well enough to where people will trust the system enough to get into this kind of trouble. But right now it seems like Tesla's in the uncanny valley of self-driving. Good enough to make people say "huh, it works" but not so good that it works 100% of the time, always, forever, no matter the conditions, other crazy drivers and all the freak occurrences that are literally 1 in a million, but given that there are 1.1 billion car trips per day in the US will happen 1100 times a day. Even at 1 in a billion it'll still happen once per day.

I'm dubious about the idea that normal people will think of airplane autopilots as autonomous. There's even a common English idiom, "on autopilot," meaning to act brainlessly.

But let's say it's true. Why would people think that, when that's not how they actually work? That's an honest question I asked above when I said, "Where did everybody get the idea that 'autopilot' implies full autonomy?"

Do we at least agree that actual factual airplane autopilots are fairly analogous to Tesla's? If so, how is it "torturing the language" to use the term that way? Overestimating the intelligence of the public when creating a term is a long way from "torture."

> Do we at least agree that actual factual airplane autopilots are fairly analogous to Tesla's?

Nope! That's the whole point of contention, actually. Autopilots are very advanced and can do almost the whole flight. Because the control problems for airplanes and cars are very, very different.

If something goes wrong with the autopilot in an airplane you've got minutes (at least) to figure it out and not die. ~30,000 feet of altitude and huge glide ratios and nearly empty skies do wonders.

In a car your average car-to-car separation in rush-hour traffic is a few seconds along the line of motion, and perhaps less sideways. If you're not on the freeway you might be a jerk of the wheel and half a second from a head-on collision.

Now you might say that the technical capabilities of both systems are the same, but nobody really cares about that. They care that it can meet some criteria that they already have in their head and they aren't going to go to school for 5 years to understand that it's not the autopilot's fault that their kid enabled in bad circumstances and died, they're going to sue Tesla to hell and back. And judges and juries will quite likely be sympathetic.

Some airplane autopilots are very advanced. Others are extremely simple. Tesla's system is analogous to something near the simpler end.

I keep talking about the simple end of what "autopilot" covers and you keep countering me by pointing out the advanced ones. I'm not denying the advanced ones exist, I'm just pointing out that the simple ones exist too, so using the word "autopilot" for simple systems is entirely appropriate.

I'm not terribly interested in the legal aspects of it, I'm just curious where everyone got this idea that "autopilot" implies autonomy (if in fact everyone does have that idea, which I'm not sure about), and I strongly disagree with your assertion that using "autopilot" to describe Tesla's system is "torturing language," given that using "autopilot" to describe a simple wing-leveling system is perfectly acceptable.

I can call a bicycle an gyroscopically-stabilized carbon-neutral human transportation device but that doesn't somehow make it fancier. It just sounds fancier.

I have argued that when people hear the word "autopilot" they think that it's very advanced. You argue that technically, something that just keeps the wings level is also an autopilot. I agree! I just disagree that the average person would know enough about the finer points of what a particular word actually means versus what just about everyone thinks it means.

Finally I would argue that if just about everyone thinks that a word means a particular thing, that's what it means. It might technically be incorrect, but words are just noises that people say to convey an idea from one brain to another. If you know what someone meant well enough to correct them, they probably used the word correctly.

Here's a major news website reporting on what an autopilot is and is not in the wake of the germanwings crash about a year ago. It maintains that autopilots on commercial jets (what people tend to think of when you say "airplane" or "flight"). This was the 3rd or 4th google hit for the phrase "what is an autopilot" http://www.cnbc.com/2015/03/26/autopilot-what-the-system-can...

That CNBC article seems to be supporting my position, not yours. It's quite clear that autopilot is an aid, not an autonomous pilot substitute, and that the pilot still handles certain parts of the flight and must be ready to take over if something happens. Unless your point is that everyone thinks autopilot means more, and this article is correcting that notion?

I fully agree that if just about everyone thinks a word means something, then it means something. I'm not trying to argue based on some technicality. What I am arguing is:

1. I'm not at all sure that this understanding of "autopilot" is as ubiquitous as you say it is.

2. Even if that understanding is nearly ubiquitous, I can hardly fault someone for using "autopilot" to mean something simpler if they are doing so because they are familiar with actual autopilot systems and are naming by analogy to the systems, not to popular understanding of the term.

3. If people really do all think that "autopilot" means a fully autonomous system, the fact that no such thing is found on real airplanes means that they're a bunch of idiots, and I'd really like to know where that idea came from.

For most people the word autopilot historically has meant full autonomy.

OK, but why? Such a thing has never existed on any production crewed aircraft.

The number of people who die every year in car accidents is so massive that any big regulatory hold-up for automated cars, which would almost without a doubt reduce deaths by at least an order of magnitude, is politically untenable.

It may be _logically_ untenable. But certainly not politically...

I don't know about US but there majority of fatal accidents are the results of deilberate misconduct, i.e. someone doing knowingly stupid things. Unless self-driving cars become mandatory they will help very little in this regards, because road hooligans won't be their target market.

One option could be for them to use the self driving cars on some routes, or maybe even on just part of the route.

If things were implemented really well, I could imagine travelling part of the journey in self driving car and then jumping to regular one for the "last mile" (or the other way round). This way one could use the self driving cars in the easy, but time consuming sections (for example congested high ways leading to airport).

Especially for the city, rain and night-time driving situations where Uber will find itself a lot of the time

As I posted in the Tesla crash thread, the current "self driving" cars are only assisting in conditions the driver actually needs the least help, good visibility and weather conditions. Its not a safety feature yet, it is little more than an advanced cruise control.

Until they can prove a self driving car can navigate a road at night in horrible weather conditions with and without other cars then its not self driving.

If anything the current crop should be emphasizing that they can warn of bad driving habits.

And, even once we get to self-driving cars, it changes the ride-sharing service's economics greatly... those self-driving cars aren't going to buy themselves.

Uber can hedge its insane market cap to buy a fleet, and then make effectively free money from that point forward selling rides.