Comment by Judgmentality | Hacker News Reader

Comment by Judgmentality | Hacker News Reader

Parent Full thread

The self-driving car play is basically an existential threat to Uber. Travis Kalanick publicly stated, multiple times, that Uber cannot work without self-driving cars. The problem is, in my opinion as someone with extensive robotics experience, self-driving cars are very far from being reality (basically everything right now is demoware).

Flying taxis are actually a lot less crazy of an idea, although it is still very difficult technology that will cost a lot of money. Kitty Hawk's Cora is going to act like a sky shuttle for tourist locations in New Zealand and plans to begin testing soon (it will probably be delayed, but the technological viability is there). There are obviously technological challenges to be solved in that space, but the biggest one is probably noise. Kitty Hawk also has the Heaviside which is very quiet (only unmanned tests so far, but it's legit). But this is a single-person short-range toy for rich people. There is also the Lilium "jet" which is an industry favorite, and the Opener Blackfly. Fun fact: Larry Page owns most of the companies I just listed and is kind of obsessed with flying cars.

People have run the numbers. You can build an eVTOL aircraft with enough range to work for something like Uber Elevate is designing. But they are ignoring noise, and I believe that is going to be the chasm they cannot cross because the public will not allow that much noise pollution (they are really fucking loud - if you live somewhere with helicopters imagine them lower, whinier, and all the time).

I think flying taxis are technologically sound but will never get approval from the public, which is actually why we don't have supersonic flight (yep - the main barrier is actually it's too fucking loud for people to allow it and many laws have been written making it illegal to fly over anything but the ocean, although there were other problems with the Concorde as well).

Uber Elevate is meant to be a carrot to convince investors they are onto something big to float the stock price. It's much more viable than self-driving cars within the next decade, but I would not count on anything happening in the US within the next 5 years apart from aircraft testing.

Blah blah blah, it's mostly smoke and mirrors.

Those human size autonomous quadcopter companies are all so optimistic about the future that they don't even realize how pie in the sky they are. Holy shit the business model on those is so crazy. Fly out to the suburbs and land autonomously in your back yard. Then it flies out and lands on the roof of your office, or at the airport, or wherever you want to go all controlled by your smartphone. Like these things are going to be flying in areas where only expert pilots would even consider in aircraft that are horrendously unstable and light and in close proximity to obstructions.

The videos always show some guy in a business suit flying some laser straight path right to the building, not being buffeted around by the wind reflected off of the buildings and one engine out from a pigeon that got sucked into the motor.

These things would have to be at least as safe as cars, and I just don't see how that's possible. They can't glide like an unpowered plane; anything goes wrong and you're immediately plummeting to your death.

At least one company includes a built-in automatically deploying parachute for whenever something goes wrong. Probably all of them will have to do that in the end.

That will help reduce fatalities but definitely won't eliminate them. Also you're gonna see a lot of people in drones strung up in trees/powerlines and in need of rescuing.

I'd be more concerned about these crashing on a street and causing a road accident.

I'm very skeptical about any widespread deployment/use of flying tech whether delivery drones or personal aircraft.

You only need to look at the opposition to runway or operating hours expansion around airports. And this is even truer if there's a perception that it's all mostly for the benefit of "the rich."

Some will argue that most people already have to tolerate road traffic today to greater or lesser degrees. That may be true but I don't see a lot of people throwing their hands up and saying "Oh well, it's already pretty noisy. Who cares about a bunch of aircraft whizzing around my house."

And you probably shouldn't understate the safety aspect and the level of pushback after the first few accidents.

I'm pretty sure the self-driving cars thing was bullshit the whole time. A great way to string along investors.

There is, as is the case with everything about Uber, no meaningful moat. There are a lot of companies and teams pouring rivers of gold into the problem, many of whom are already profitable from other businesses and who can raise debt at relatively low cost (vs diluting shareholders again, and again, and again).

The carmakers will have their own services. Google will. Everyone else who rounds up a few shekels from friends and family will do it in their hometown. The taxi companies will, so will the livery companies. Car financiers will pile in with monthly payment schemes "but with an app!". There'll be ten "AirBNB for Uber" startups enthusiastically shanking each other with VC shivs.

When the dust settles it will at best be a game won by those who were already profitable coming in, at worst a game of musical bankruptcies.

Nothing about self-driving cars will save Uber. Only raising their price to cover the cost of every ride and accepting a dramatic reduction in topline revenue will do so.

> I'm pretty sure the self-driving cars thing was bullshit the whole time. A great way to string along investors.

I somewhat agree that self-driving cars isn't an obvious solution for Uber when you think about it, because it basically makes the entire business of Uber a stepping stone towards a completely different company. That is to say the majority of everything Uber has built would no longer be relevant. Of course, I suspect Travis always knew this and really it was actually Uber itself that was always bullshit (or at least by the time the company became a unicorn, nobody really knew anything when they were still a glorified limo company) and self-driving cars always the real endgame. The problem was it didn't go according to plan.

Considering the relationship Travis and Anthony Levandowky established and the whole history there (Uber legitimately tried to defend Anthony from the Waymo lawsuit until it was blindingly obvious they could no longer do so), I think Travis was dead serious about self-driving cars for a long time. The problem is Travis is a great entrepreneur and businessman, but robotics is an unforgiving field.

It's easy to sell an idea. "Wouldn't self-driving cars be great?" And honestly 5 years ago I felt like it might be possible, and many other seasoned roboticists legitimately got on-board (so not just for the money, although some did and I don't blame them because holy shit I have seen some of those offers and they were insane). But at this point, and this is very much an open debate, I believe you need to solve general AI before you can have a meaningful self-driving car product.

Everything above applies for L4 autonomy, anything that requires driver attention (such as Tesla's Autopilot) or extremely limited use-cases (such as GM's SuperCruise, which is 100% independent from GM's self-driving car division called Cruise) is a completely different discussion.

I briefly bought into the self-driving hype too, but I had a useful reference case to help me come down to earth: self-driving trains. Rio Tinto (the world's second-largest mining company) has sunk a lot of money into developing self-driving trains.

Trains. That have right of way. Their own trains, that they own, on their own rail, that they own, liberally covered with communications towers and sensors (many of which my father put up during his time there).

It's about as close to the simplest version of self-driving there is. And they weren't hurting for smarts: they raided quite a few universities for even vaguely relevant talent.

They did it, though. Level 4 autonomous trains. It cost them a billion dollars and six years just for one crack at the software. Before that at least a decade of work for who knows how much money.

But: trains. That's the functional state of the art worldwide. Trains.

I would love some links to this case. That is a wonderful counter to the endless reams of self-driving car propaganda that shows up in my personal life on a frequent basis.

Recently saying that they had successfully tested it: https://www.railwaygazette.com/australasia/rio-tinto-complet...

Earlier announcement (2014) that they were planning to do so: https://www.railjournal.com/signalling/rio-tinto-gets-set-fo...

Note that the original contract with Ansaldo for development was approximately half what it wound up being. That they will admit to. And I suspect leaves out a lot of the work around the lines which can be used for autonomous and non-autonomous purposes.

But then it goes back further, to this article from 2008: https://www.railwaygazette.com/news/rio-tinto-to-go-driverle...

Predicting that autonomous trains would be done in 5 years. That is, they expected to be finished by 2013 ... when the next contract was signed.

I won't be surprised if it's announced a few more times in the coming years.

Mine automation is a big deal, Rio Tinto have spent heavily on it. Not just trains, they also have spent a lot on developing autonomous mine trucks. The pit is covered with a lot of beacons to help the trucks position themselves and they use regular scans of the site to update the map. I'm not sure how far they got with those.

BHP-Billiton seems not to have gone as far, but they have spent a lot of money in remote operations: having the control centres for multiple mines consolidated into special-purpose offices in Perth etc. Cameras everywhere, das blinkenlights, the whole shebang.

All of these offerings from Uber appear to exist solely for pumping up the stock price. The flying taxis thing is, quite obviously, not going to happen for the reasons you say, and a number of other reasons (proof of flight worthiness will require very significant testing time).

It's not really worth taking any of these ideas seriously. All they illustrate is that Uber doesn't believe their core business is viable.

Nice analysis!

But if it’s really just too loud, isn’t there some way to muffle the sound? I suspect it would be expensive but certainly not ridiculously so.

The answer is actually no, you can't muffle the sound without ruining the aerodynamics that allow it to fly. Or at least, nobody knows how.

What would it take, smooth out the pressure waves into white noise from a continuous stream of air?

It doesn't matter how "white" the noise is, what matters is the volume. A jet engine is pretty close to white noise, it's just really fucking loud. We accept jets because they usually are 5+ miles above us. These flying cars are meant to fly a few hundred feet above the ground.

My hunch, although nobody can say definitively, is it is physically impossible to make these quiet enough to be acceptable for mass use. But aerodynamics has as much trial-and-error as it does theory so maybe a genius will come up with something novel.

Reply on news.ycombinator.com