Slashdot Mirror
Google Unveils New Self-Driving Car Prototype
colinneagle writes In May, Google released a teaser image showing a mock-up of the autonomous vehicle it planned to build. Today, the company followed up with an image showing the finished product. Google says the first edition of its self-made self-driving car will feature "temporary manual controls as needed while we continue to test and learn." When Google introduced its prototype back in May, the company claimed its self-driving cars "won't have a steering wheel, accelerator pad, or brake pedal because they don't need them." Apparently, it still has yet to reach that point. The development is an important step forward for Google's driverless car efforts, which have been deemed impractical by many of late. Last year, the Financial Times reported that Google had difficulty finding manufacturing partners that would build vehicles featuring the self-driving capabilities used in its Prius. In that light, maybe Google's willingness to build its own hardware just to get the technology on the road means that its self-driving car team knows something the rest of the industry doesn't."
The Oatmeal posted a review of the car and state of Google's technology in general:
http://theoatmeal.com/blog/goo...
It doesn't hurt to be nice.
Actually, the controls were added because of a California state law requirement for them, not because Goggle thought they were necessary...
It looks like it is designed for indoor, otherwise it is too dangerous to drive it because other drivers can just miss it :)
Why Google? WHY? Why does every manufacturer of cutting edge vehicles, like EVs, have to make them so damned ugly? Why can't we get a car flavored car?
How do these self-driving cars handle situations where, say, the power goes out and you've got a cop in the intersection waving traffic through, holding their arms out to stop, okay you three can go, etc. Or where a human schoolbus driver extends the stop sign as it slows down... or any situation that requires a kind of interaction between driver/pedestrian (construction worker saying hold up, etc.) or pulling over and to the right because there's an ambulance coming down the road on the opposite side of the street...
Obviously these situations have been considered.. but have they been solved?
So if I buy a production Google car without controls, and the vehicle runs off the road and kills 20 kids in a playground, who is at fault?
A. Me (for owning it)?
B. Google (for shoddy programming)?
C. The state (for allowing driverless cars)?
D. The Kids (because they should have gotten out of the way)?
Related question - do I need to carry insurance to use one of these cars as I am not driving, I'm merely riding?
Essentially, what is the model for this type of vehicle - am I the 'driver', or am I 'passenger' (like on a bus or taxi)?
If they can sort this out maybe I'll buy one someday.
Left MS Windows for Linux Mint and never looked back!
Vote for Bernie in 2016!
>> maybe Google's willingness to build its own hardware just to get the technology on the road means that its self-driving car team knows something the rest of the industry doesn't
Or...that the rest of the auto industry doesn't want to get tagged with the "first death caused by an automated car."
In that light, maybe Google's willingness to build its own hardware just to get the technology on the road means that its self-driving car team knows something the rest of the industry doesn't.
Unlikely - they would have shared what they knew with the people they were trying to partner with (subject to patents, ndas etc). More likely the car makers just don't share Google's appetite for risk in advancing unproven technology. And maybe they're right - it's the second mouse that gets the cheese.
Or maybe the rest of the industry isn't willing to risk the liability for when something goes horribly wrong. Fear of litigation is a real impediment to innovation. But in this case, there is a huge amount of risk. Dealing with the real world happing around you while you're trying to make the computer drive the car has a ton of non-trivial potential disasters waiting to be exposed.
Marissa Mayer's 2011 concept car sketch for BMW vs. Google's 2014 built vehicle
Looks like the same people who designed Google+ were responsible for this inbred result of a car and a toaster. Surely one mechanical engineer on the team uttered the word "aerodynamics" somewhere along the way?
No. I fully believe that the cars HAVE reached that point.
The thing is, many people (including legislators and insurance underwriters) don't trust such a system yet. Thus, the car has conventional controls as a "failover to manual" in cases of catastrophic systems failure.
Honestly, while I believe that you can build self-driving cars. And that they can be safer. *I*, personally, don't want one.
Put simply, I refuse to relinquish that level of control over my driving experience. Ever.
Well, maybe when I'm in my 80's or something. By that point, I'll have little left to lose.
Chas - The one, the only.
THANK GOD!!!
so many people were outraged at the thought of not having manual controls that the rules were changed to require them.
That's why this version has the temporary manual controls. Google is hoping to get the rules changed
Realistically, if the car has been driving on auto with the passangers not paying constant attention (in which case, why do you have the auto controls? and people are really bad at paying constant attention to something that they can't control), do you really think that having someone grab for the wheel in an emergancy is going to do more good than harm? the big red panic button to stop the car is about all that is meanignful in any case.
Remember that these cars are 'Community Electric Vehicles' aka glorified golf carts, with a top speed of 25 mph, because the rules for producing such vehicles are so much smaller than the rules for produceing full capacity vehicles.
David Lang
They look completely different other than the rounded edges. What's your point?
A driverless car with no pedals and steering wheel doesn't need to be like a normal car. Where's the sleeping area? Where's the flip-down table to work on? I don't want to stare at the road if I'm not allowed to drive.
But I wouldn't buy one anyway. I always want brakes. I'd prefer a steering wheel so I can do stuff that the AI won't let me do, like drive through my backward to drop off heavy things in the garden, etc...
Realistically, if the car has been driving on auto with the passangers not paying constant attention (in which case, why do you have the auto controls? and people are really bad at paying constant attention to something that they can't control), do you really think that having someone grab for the wheel in an emergancy is going to do more good than harm? the big red panic button to stop the car is about all that is meanignful in any case.
The situation they require manual controls for is when you drive into a blizzard/flood, and the car drives until it's unsafe to stop and unsafe to continue. When you give the computer the choice between two bad things, something bad will happen. The regulators would rather that bad choice be in the hands of a human, when the "fix" is to work out as many "unsafe to continue, unsafe to stop" conditions, and improve stopping before them. It's not about the instant hand-over, but the impossible situation.
Learn to love Alaska
No. I fully believe that the cars HAVE reached that point.
It does not matter what you believe we factually have not reached that point yet. Google themselves have admitted that their vehicle does not work well in rain, snow or fog. There is also the issue that every road would have to be scanned and manually gone over for current technology cars to drive on them. I have yet to see a valid test of a vehicle driving into a random, unaugmented parking lot and parking.
Just last week I encountered a cop with a lighted baton who was directing traffic from the side of the road. He would stop traffic, walk to the middle of the road while motioning people across the road with his baton, then walk off the road while waving the baton *behind his back* to signal "go ahead".
Does the self-driving car recognize this sort of thing?
Will it drive when there's snow on the ground?
I think I'd keep the steering wheel and manual control - just in case..
More here: Is Google CEO's "Tiny Bubble Car" Yahoo CEO's "Little Bubble Car"?. Thought it was interesting that the no-frills bubble car Google came out with in 2014 was closer to Mayer's 2011 vision, and quite a departure from the modded Priuses and Lexuses that Google had shown off in the past. Your mileage may vary. :-)
Oh my god that thing is so ugly. Just another example showing that Google is great at engineering, but terrible at user interfaces.
Who said emergency? An emergency is probably exactly when you want a computer to be in control, simply because it can process more information more quickly, and the decisions to be made are trivial and minimal (aka "bring vehicle to a safe stop, right now").
But I would want manual controls on my car of the future because on some weekends I drive into the countryside and I drive on small dirt roads that may or may not be on the map. Or to festivals or other big events where at the end you park on a field. Or you drive through a really crowded street where the computer will most likely just stop and stand because there's always someone in front of the car.
There are plenty of non-emergency situations that I'm not sure the automatic driver can handle.
Assorted stuff I do sometimes: Lemuria.org
Whilst she may be Yahoo's CEO, back in 2011 she was working at Google.
The situation they require manual controls for is when you drive into a blizzard/flood, and the car drives until it's unsafe to stop and unsafe to continue.
I can imagine that going over so well with consumers "Hi! It's me, your autonomous car here. You know how I drove you up in the mountains and to this mountain pass? Well now there's a blizzard coming so I quit. Now I know you haven't touched the wheel in a month because I've been doing your commute and I wouldn't drive under these conditions, but you'll probably freeze to death if you don't get down so... best of luck? Toodeloo."
Live today, because you never know what tomorrow brings
Very interested to see the legal forms the buyers will have to sign which covers googles liability on these things.
People are very creative in thwarting logic of machines and do many unexpected things that these things logic very likely wont handle very well.
>>When Google introduced its prototype back in May, the company claimed its self-driving cars "won't have a steering wheel, accelerator pad, or brake pedal because they don't need them."
This is what frustrates me most about Google products. They are always "improving" their interface, which means adding automation and taking away simple ways to interact directly with the application. Their Contacts list for example--I've gotten used to the quirky way it works now, but I do miss the simple version that used to be available. That was still a whole lot fewer clicks and faster to add contacts. In my OPINION, the google UI designers are very poor at functional UI design, though they are certainly innovative at obfuscating previously simple controls behind lots of javascript.
People (justifiably) rail on Microsoft for things like removing the start menu, and creating a super-secret hidden charms bar that only shows up if you happen to know where to hover your mouse. Google is nearly as bad.
It is true that the current prototypes cannot handle some extreme situations.
However, it seems like we're getting close (from a tech standpoint) to something that could operate in a limited but useful capacity: operating like a taxi in an urban area in a place where it doesn't rain much or snow at all.
There are places like this. Los Angeles and San Francisco, for example.
Choice of location limits exposure to extreme weather conditions. Simply stop providing the "taxi" service in the rare case where there is heavy rain, and make the congestion resulting from the resulting mode-shifting someone else's problem. (This would be a business, so no need to be considerate to anyone else, as Uber has shown.)
That then leaves the construction or police rerouting problem. If the self-driving cars are operating as a taxi, they are presumably administered from a central location that can be informed of (or discover) disruptions and disperse that information to the self-driving cars so that the cars can avoid encountering these problems. In San Francisco there are few places that can't be reached by an alternative route, and cul-de-sacs can be handled by getting you to the closest intersection and having you walk a block or two.
There will likely be exceptional cases where a disruption arises quickly and the self-driving taxi system is unable to adapt in time. In this case, we hit the limit of today's system and the best it can probably do is try to stop at a safe location and call for help. If it were one car and it made a good decision on where to stop then this could be acceptable but non-ideal. If it were a whole fleet of them then they could end up blocking a street where emergency vehicles need to pass, which would obviously be bothersome and an area where improvement is needed before mass deployment.
Thats' why the GPS the car uses should take weather and mountain passes into consideration, and refuse to leave the garage if the weather conditions outside are unsafe at any speed.
The only failure I've ever seen anyone focus on with self-driving cars is where the user makes multiple errors "urging" a car into a known unsafe area. The "fix" to that is to remove more power from the driver. If "I'm sorry Dave, I'm afraid I can't do that." were the response to all the stupid human requests, then 99% of complaints against driverless cars would be invalid. Is that really the reality you want to push for?
Learn to love Alaska