Support NeoGAF

[Joco]

This is going to put so many people out of work. And politicians and the press aren't talking about it.

At least this should get the conversation rolling on how society deals with the human aspect of work becoming obsolete. Because machines and AI are going to put millions out of work in the coming decades.

 

[7DollarHagane]

In the US, driving a truck is the most popular job in most states.

http://www.npr.org/sections/money/2015/02/05/382664837/map-the-most-common-job-in-every-state

I think that semi trucks will still need manned operators but it will be more like a train operator, where they are there to watch for malfunctions or perform manual tasks with the vehicle. This will make their job a lot easier. With a vehicle as large and unwieldy and deadly as a semi with trailer I don't think they should ever go full unmanned with them.

 

[sp3ctr3]

I feel sorry for the bus-, taxi-, truck- etc drivers who will be out of work when this truly kicks in.

I enjoy riding my car and sometimes go to the track, but man, I have a 45min commute each way to work and being able to just get in my car, kick back with a cup of coffee and play a handheld or take a nap sounds pretty awesome.
Also, no designated driver when out drinking

 

[Calabi]

This tech actually isn't as far along as I'd believed it was. Google's system fucks up once every 74,000 miles? I'm a way better driver. (~500,000 mile sample size)

Google isnt neccesarily the leader though just the most publicized one. There making stupid bubble cars that no one is going to want whilst all the conventional car manufactures are leveraging there existing cars.

Google doesnt have all the data, it doesnt have accurate maps. The car manufacturers are making and selling cars now which have all the cameras and systems that can gather that data, or contracting from others.

It's going to make for a much smoother and inevitable transition if the car you buy to drive is able to take over more and more of the driving down the years.

 

[anonymous_abc]

Google isnt neccesarily the leader though just the most publicized one. There making stupid bubble cars that no one is going to want whilst all the conventional car manufactures are leveraging there existing cars.

Google doesnt have all the data, it doesnt have accurate maps. The car manufacturers are making and selling cars now which have all the cameras and systems that can gather that data, or contracting from others.

It's going to make for a much smoother and inevitable transition if the car you buy to drive is able to take over more and more of the driving down the years.

Google doesn't want to sell cars, they want to sell their services to other car manufacturers!

 

[Joni]

I want to see how they do on European roads. American roads seem like the Easy difficulty, you go for Europe for the medium difficulty with Belgium as the difficult boss spike. If it can survive that, they can move onto the High difficulty in India.

 

[Angry Grimace]

Skynet gonna make cars drive into the ocean

 

[Kenshin001]

I really don't care about autonomous cars at all. I really wish the car industry would hurry up and shift to electric vehicles. That is far more important.

 

[anonymous_abc]

I really don't care about autonomous cars at all. I really wish the car industry would hurry up and shift to electric vehicles. That is far more important.

What makes you think they are mutually exclusive?

 

[Calabi]

Google doesn't want to sell cars, they want to sell their services to other car manufacturers!

Well I hadn't heard that, what are they selling exactly?

Because, the maps arent up to date, and the cars systems are expensive(cheaper are available). Why would the car manufactures buy off of Google?

 

[Nipo]

Well I hadn't heard that, what are they selling exactly?

Because, the maps arent up to date, and the cars systems are expensive(cheaper are available). Why would the car manufactures buy off of Google?

Their algorithm has by far the best accuracy and least human interventions according to the wapo report. It wasn't even close.

 

[Calabi]

Their algorithm has by far the best accuracy and least human interventions according to the wapo report. It wasn't even close.

So there going to all that work just to sell software? And all the car manufacturers are going to buy it because they arent going to be able to do as good a job given enough time.

 

[yuraya]

Just imagine being on a 6 lane highway doing 70 with semi trucks (plus other cars) all around you and not having control of your car. Having some computer drive for you. Putting your life in the hands of these tech companies.

I would probably have an anxiety attack and shit my pants. There is no way this will happen in 4 years time. Fuck off and fuck the Demolition Man future.

 

[Nipo]

So there going to all that work just to sell software? And all the car manufacturers are going to buy it because they arent going to be able to do as good a job given enough time.

Exactly. Same reason it is cheaper for oems to just pay for Windows or use android than try to develop something better. Development cost and patent licensing can cost billions. And "powered by Google" holds more weight with consumers than "built by GM"

 

[anonymous_abc]

Just imagine being on a 6 lane highway doing 70 with semi trucks (plus other cars) all around you and not having control of your car. Having some computer drive for you. Putting your life in the hands of these tech companies.

I would probably have an anxiety attack and shit my pants. There is no way this will happen in 4 years time. Fuck off and fuck the Demolition Man future.

So I take it you don't fly then?

 

[SolarPowered]

Just imagine being on a 6 lane highway doing 70 with semi trucks (plus other cars) all around you and not having control of your car. Having some computer drive for you. Putting your life in the hands of these tech companies.

I would probably have an anxiety attack and shit my pants. There is no way this will happen in 4 years time. Fuck off and fuck the Demolition Man future.

lol

Commercial transport will become automated far faster than the regular car owner population. I bet the streams will cross when regular people are the ones being sued by companies because human drivers inevitably falter before our superior AI overlords on the roads.

 

[runningjoke]

Just imagine being on a 6 lane highway doing 70 with semi trucks (plus other cars) all around you and not having control of your car. Having some computer drive for you. Putting your life in the hands of these tech companies.

I would probably have an anxiety attack and shit my pants. There is no way this will happen in 4 years time. Fuck off and fuck the Demolition Man future.

But you can get drunk before getting in. I know I would

 

[ClosingADoor]

I want to see how they do on European roads. American roads seem like the Easy difficulty, you go for Europe for the medium difficulty with Belgium as the difficult boss spike. If it can survive that, they can move onto the High difficulty in India.

Pretty much. Highway driving is fine, but I can't see them performing that well yet with a ton of pedestrians and cyclists around making sudden movements.

Also wonder how this performs when a road is filled with snow.

 

[Foffy]

The sooner, the better.

Especially if they're disruptive as they are very likely posed to be. I can't wait.

 

[Des0lar]

Pretty much. Highway driving is fine, but I can't see them performing that well yet with a ton of pedestrians and cyclists around making sudden movements.

Also wonder how this performs when a road is filled with snow.

Nvidia just recently showed their technology working great in congested cities, with smart learning systems, which even proved working in snowy conditions. This was now, on unoptimized software. Give it 3-4 years and that stuff will just work.

 

[hodgy100]

Ugh, this is awful. The only time I would like it is if it could drive me home when I'm drunk. But that's what Uber is for..

I would hate not owning my own car, and having to share with random people, that's just grossé.

who said you would have to share?

 

[PdotMichael]

lol

Commercial transport will become automated far faster than the regular car owner population. I bet the streams will cross when regular people are the ones being sued by companies because human drivers inevitably falter before our superior AI overlords on the roads.

Well, people aren't buying significant more expensive cars just to automate their average 45min of daily driving. The transport industry is way more interested in running vehicles which can basically operate for 24h every day.

 

[Foffy]

it's incredible how big this industry is. the roll out of autonomous vehicles will probably take forever, especially to commercial trucking, but i wonder where the people will drift off to? it'll be something that they have to deal with at some point in their lives, even if it's not till 2030

Would it shock you that such a thing has been automated and been on the roads for almost a year? We already are accomplishing this.

The trigger to automation will the cost and effectiveness of the transition, and this is always a game of when, not if.

 

[TheOfficeMut]

I haven't really kept up on the technology, but how can they be aware of how much to turn on high ways in order to remain in lane, especially if the traffic likes have been worn out and aren't visible.

 

[felipeko]

However, fully autonomous cars as people imagine it today are vaporware right now. Hardware for 360 monitoring that can fit a car without huge sensor on top does not exist today. They are hoping it will in 3-4 years, but it doesnt today.

Google is advertising their tech through blogs but if you believed them until few weeks ago, they were saying that they never had a problem with their system. Now with state legislature making them reveal number of times driver had to take over, it is suddenly a lot more. Even then, they are making vague statements.

Here is TIME's article:


http://time.com/4179166/google-cars-self-driving-autonomous/


So one "problem" per 5318 miles. With majority of those coming from city driving, which likely means a lot more problems than that in city driving.

Now imagine that with New York traffic or any European city traffic vs freaking Mountain View.

So, who cares if there's a big sensor on top of the car? No one is going to buy it anyway, people will just call Uber/Lyft and they will not worry about cars being ugly. It's like people saying "who the hell will talk on a 5'' cellphone?"

About Google's problems, not crashes, i'm pretty sure Google is aware that we are not on 2020 (when they plan to release the car), so they are still perfecting the technology. It got 6 times better last year, and will probably keep doing that for the next four years (not unreasonable with better AI and vision), and so the car will have a "problem" for every 6 million miles. Much better than any human driver i would argue, and that's all they have to beat to be fully autonomous.

 

[nOoblet16]

Guess if it doesn't work with rain it'll be useless in UK.

I do know one country where it will take 50 years rather than 5 years though, simply because of how traffic works there. It'll be a long time before people start following driving etiquette, considering even with autonomous cars there will be plenty of people who drive on their own. But this is also one country where a fully autonomous traffic will have massive benefits !

 

[Kenshin001]

What makes you think they are mutually exclusive?

I don't. I just think that fully autonomous cars will come before all cars are electric. The car industry is too heavily invested in gasoline power. I see car makers like Mercedes showing off the autonomous ability of their cars but there is no mention of an electric version.

 

[Alexlf]

I haven't really kept up on the technology, but how can they be aware of how much to turn on high ways in order to remain in lane, especially if the traffic likes have been worn out and aren't visible.

Distance from side of road/other cars/stored road data/etc. How do you think humans do it, magic?

 

[Gold Coins for Cats]

I don't. I just think that fully autonomous cars will come before all cars are electric. The car industry is too heavily invested in gasoline power. I see car makers like Mercedes showing off the autonomous ability of their cars but there is no mention of an electric version.

Putting aside the fact that battery technology is unable to charge fast enough to equate the time it takes for one to fuel up their car, you still need to manufacture the electricity, and the caustic batteries, which take up huge 'carbon footprints'.

There obviously can be money changing hands to stall growth, but it's not the only issue by a longshot. We've had 90+ years to refine gasoline engines, as opposed to the 15 for electric cars.

Edit: I know I said I was going to put aside the first point, but I was reminded of this article and it was just too funny: http://www.telegraph.co.uk/news/worldnews/northamerica/usa/10592660/Charge-Rage-electric-car-owners-get-angry-after-having-vehicles-unplugged.html

 

[anonymous_abc]

I don't. I just think that fully autonomous cars will come before all cars are electric. The car industry is too heavily invested in gasoline power. I see car makers like Mercedes showing off the autonomous ability of their cars but there is no mention of an electric version.

If it eases your mind fully electric cars will come before all cars are autonomous.

 

[Kinitari]

Google isnt neccesarily the leader though just the most publicized one. There making stupid bubble cars that no one is going to want whilst all the conventional car manufactures are leveraging there existing cars.

Google doesnt have all the data, it doesnt have accurate maps. The car manufacturers are making and selling cars now which have all the cameras and systems that can gather that data, or contracting from others.

It's going to make for a much smoother and inevitable transition if the car you buy to drive is able to take over more and more of the driving down the years.

Google is the most likely furthest along. They have the most senior team and have the most data. It also has the most accurate maps, as they only drive on pre 3d mapped roads. Their once per 74000 miles disengagement is for the entire life of their testing, they've had recent spurts of no issues for 130k+ miles. And this isn't about incidents, like crashes, it's about disengagements, which are moments the cars feel are beyond their Ken and switch to manual. Considering the rate of progress, I wouldn't be surprised if that length of time doubles this year - the only reason I see it not doubling is because they're going to be testing it in harsher conditions - snow, ice, etc.

Their goal is to have entirely autonomous cars in around 5 years, and most likely license out their tech to other car manufacturers, like Ford, who they recently partnered up with.

 

[Kinitari]

They will be "drivers assistant" for a decade or so at least.

Trully autonomous driving from your home to park at work is never going to be properly possible until we get smart roads which have sensors that tell the car computer what is going on. Japan just started implementing those, but it is very slow roll out.

But in the meantime, you will have cheap pre-crash in every vehicle that will make everything much safer.

This tech is not expensive - Toyota managed to bring cost of pre-crash down to $300 to customer for affordable yet good system (most right now are not good, for instance German companies have very poor performing pre-crash systems in real life). Toyota's best system costs $600 and has radar cruise control and bunch of other things.

So if that can be sold at $600 for profit, then in 6-7 years we will have same system doing "assistance" driving on its own for $1k or less.

However, fully autonomous cars as people imagine it today are vaporware right now. Hardware for 360 monitoring that can fit a car without huge sensor on top does not exist today. They are hoping it will in 3-4 years, but it doesnt today.

Google is advertising their tech through blogs but if you believed them until few weeks ago, they were saying that they never had a problem with their system. Now with state legislature making them reveal number of times driver had to take over, it is suddenly a lot more. Even then, they are making vague statements.

Here is TIME's article:


http://time.com/4179166/google-cars-self-driving-autonomous/


So one "problem" per 5318 miles. With majority of those coming from city driving, which likely means a lot more problems than that in city driving.

Now imagine that with New York traffic or any European city traffic vs freaking Mountain View.

The solid lidar vs spinning lidar exists today, and has already been used in demos, and should be available to manufacturers in 2017, for 250 bucks a pop - significantly cheaper. The first generation of that tech looks more like a small block, but you know... First gen by next year is pretty good.

Google, and others, are fighting hard for 100% autonomous driving, and I think they'll get it. The issues that they highlight the cars having are almost all people rear ending them - and they've been very open and honest about what they're doing to get that number down more and more - they haven't had an incident in two months I think, a large part from their improvements - even though they've increased the number of cars on the road.