Hacker News Reader

Comment by adamwong246 | Hacker News Reader

huzzah! Another step closer to the dream of fully autonomous automobiles! Though human uber drivers must know that they are, at max, 5 years away from obsolescence?

View on HN

"human uber drivers must know that they are, at max, 5 years away from obsolescence"

Care to place a friendly wager on that? I'll even forgive the "uber drivers must know" part of your crazy statement since that obviously isn't correct.

Ubiquitous driverless cars are coming. 5 years is an incredibly aggressive time frame.

It's a ridiculous timeframe if you consider that there are no driverless cars for sale right now. Even if Tesla or Ford or any other vendor started selling them tomorrow, the driverless cars would be too expensive and the factories could hardly pump out enough of them to put every Uber driver out of a job in just 5 years.

I'm looking forward to being able to order a driverless car from my phone, but in just 5 years, these cars might still be scarce, illegal in most places and they might suffer from limitations in their driving ability that prove problematic.

You're thinking about it wrong. No consumer is going to buy a driverless car. Instead you'll call an Uber or GoogleTaxi and there won't be a human behind the wheel. Just like Musk said, owning s car is going to become like owning a horse. People do it as a hobby but not as a necessity.

I agree. We'll buy into car share networks. There will be super primo services where you can ride in a limo, business class networks that allow you to ride in privacy and maybe even economy class networks where busses and vans carpool like uber's new hop feature. Crucially, this allows the number of overall cars to slashed. 1 auto-auto can do the work of 10 non-auto-autos, and the cost will be spread out among subscribers and expensive electric cars will finally be cheap enough for everyone.

Driverless cars don't have to be custom-built: you can add a kit to an existing car that contains a Linux box, LIDAR, cameras, and appropriate machinery to drive the car's controls. Most of the existing Google prototypes have been Priuses or Lexuses with the kit. IIRC, all the components are pretty inexpensive; by far the biggest cost is the LIDAR, but that's because they're not produced in volume and so you have to pay custom-built prices.

Ok, maybe it's too optimistic but I'd totally make a wager. I say it will take 5 years for the Bay, 10 for major cities and highways, 15 for the rest of the country and by 2050, human driving will be illegal on public roads.

any takers?

To be clear: you want to bet that in 5 years human driving will be illegal on public roads in "the Bay"?

no, I don't think we'll outlaw human driving for several decades. It will take a while for the old holdouts to die off- those individuals who'll stop driving when you pry the steering wheel from their "cold dead hands."

But I do think that the legalization of auto-automobiles will start in San Fransisco and the surrounding area.

Probably fewer and fewer people will go to the effort of learning to drive and passing the test, particularly in places where it's difficult and expensive.

I don't think so. Rather, by 2050, driving will have been made illegal except on closed courses.

Personally I don't think it will ever have to be illegal. It's more likely that people will stop driving of their own volition and the laws allowing driving will remain in place, just be unused.

I'm actually in agreement that autonomous transportation will become "a thing" in roughly five years, somewhere in the world, in some capacity. This will prove the model and eliminate most of the rational arguments for its prohibition, which will lead to it becoming commonplace but probably over the course of several decades.

If I had any money, I'd probably take you up on that. I think the technology might be ready in that time frame, but bureaucracy and safety concerns will push it back. We don't know how well the vision/detection methods of driver-less cars will handle certain scenarios, like a solar flare, or direct subversion from other humans. Would someone be able to disable my car from moving and rob me with a few small devices?

I agree that bureaucracy is the main hurdle. The technology is so close but I'm not sure that the populace or the government is ready.

But malicious agents can disable your car today. A solar flare would fry you current car's electronics just as easily as an auto-auto. And a gun qualifies as a "small device" too, so I'm not sure why that makes in difference.

"Self driving" cars which require a human to be ready to take over are totally plausible in that time frame...

But are you suggesting that in 5 years a person can summon an autonomous taxi in the Bay area? By autonomous car I mean the client (passenger) will not need to be ready to take control nor will there be any other human in the vehicle for that purpose, and the vehicle will transport the client from one place in the Bay to another over the public road system, mixing with other non-autonomous traffic of various types.

If so, I'm in for USD $100. I say that in 5 years (12/17/2020 or before) it will NOT be the case that a user can expect to summon an autonomous (as defined above) taxi for transport anywhere in the United States.

Ok, consider that wager taken.

I think the key parts of the wager are 1) that the car drives itself, without a human present 2) it's in the Bay area and 3) it's legal.

I'll see you in 5 years.

This might be a ridiculous question.

But, why don't we have self flying pilotless airplanes yet? I'd imagine problem set is much more limited and tons of research has already been done.

> But, why don't we have self flying pilotless airplanes yet? I'd imagine problem set is much more limited and tons of research has already been done.

We do, but they're mainly used to surveil people instead of move them around.

I think for the most part, we do. The pilots are part of the "theater of security"- the computer does most the work. Many people wouldn't ride in planes without a human behind the wheel. And this fear of machines is the biggest hurdle to auto-autos.

why do autopilots keep turning off every time the planes run into problems? like the air france and the air asia indonesia crash turning control over to pilots is a recipe for disaster.

They do eg when there is a mismatch between the numbers they getting from 2 different sensors.

But the fact that many more people will work for uber will make shifting to autonomous cars even more difficult.

Reply on news.ycombinator.com