Google's Driverless Cars Now Rolling In the Heart of Texas 114
MarkWhittington notes that, as reported by The Wall Street Journal, Google has started testing its self-driving cars in Austin. These driverless cars, loaded with the sensors, GPS transponders, and cameras, are now in service in "an area northeast and north of downtown Austin. The purpose of the test drives is to see if the car's software works in driving conditions outside of California and to develop a detailed map of Austin city streets. Each self-driving car has two human drivers ready to assume manual control if something goes wrong."
Question (Score:5, Funny)
Re: Question (Score:2)
More than it takes to drive a non-driverless car, apparently.
One step forward, two steps back? Lol.
Re: (Score:3)
On mopac during rush hour? It takes none. You turn the car off, get out, lay in the median and try to wake up before 5pm so you can turn around and go home.
Re: (Score:2)
Re: (Score:1)
Next question: how many standby drivers should one have on-hand while testing prototype driverless cars?
None. Because we should no more allow prototype cars on public roads than we should allow prototype planes on public airports. They have to be tested to far beyond the prototype stage before being admitted. If a company isn't confident enough that it can be let loose in real world situations, it isn't ready yet.
Re:Question (Score:5, Insightful)
You realize prototype and home built airplanes are flown from public airports all the time?
Re: (Score:2)
I should, perhaps, have said "public big city airports". You don't see many prototype aircraft at JFK.
As for home built planes, they have to be approved before flying them. Not after the fact. A big difference.
Re: (Score:2)
Prototype planes are registered as "experimental aircraft". That means that the authorities have looked at it and decided it's safe enough for a test pilot to fly it. Proper type approval comes later when the manufacturer has gathered enough evidence by (among other things) actually flying the plane.
In the USA home built aircraft are also registered as experimental aircraft (despite not being truely "experimental" in most cases) and get much the same level of scrutiny. Other countries may have different rul
Re: (Score:2)
Re: (Score:2)
Of course we do.... where else are we going to launch and land large prototype and home-built airplanes?
Your neighbor's backyard? Close down the nearest highway and use the road as a runway?
Public airport beats the alternatives.
Re: (Score:2)
Are Texan Drivers worse than Californian Drivers?
Disclaimer: I've not from the USA and I've only ever driven one time in San Jose.
(apparently I shouldn't have driven through the neighborhood with boarded up windows...my bad, I was lost!)
Re: (Score:2)
In a word, yes. I have lived in both, and in austin I have seen people going the wrong way in the single lane under an overpass to change directions. I have no idea how they even got there. I have seen people in a 3 lane wide one way street make a left turn (in front of me) from the rightmost lane of the road. I was in the middle lane. I routinely watch people run red complete red lights, not pink, but steady red's. Staying within the lane of multi-lane roads also seems to challenge drivers in austin. I hav
Re: (Score:2)
Having driven extensively in both states, I'll say the drivers in Texas are more clueless and likely to do stupid things by mistake. The drivers in California aren't necessarily bad, but they are much more aggressive and more likely to do something stupid out of impatience.
As always, YMMV on any particular drive in either state, this is just my general impression over several hundred thousand miles of driving.
Re: (Score:2)
California drivers never heard of lane discipline. They think nothing of (as Denis Leary might put it) "driving really slow in the ultra-fast lane," or of passing the aforementioned assholes on the right.
Re: (Score:3)
Are Texan Drivers worse than Californian Drivers?
Given that the most popular car in TX is the F-150, there are 85 mph speed limit zones, and that car dealers offer to throw in the (not "a") gun rack for free, what do you think?
But anyhow, this isn't testing the car in Texas, it's testing it in one fairly well-regulated city, atypically designed to have as many broad streets, identical size city blocks and 90 degree corners as possible. It didn't evolve like most other cities, and present fewer challenges except for traffic density. Now Houston would hav
Re: (Score:2)
A better challenge would be pretty much any city in Europe.
Re: (Score:2)
I don't get your argument. Perhaps you are letting your political view get in the way of rational thought.
F-150 Pickup truck. How does that make bad drivers? They have a higher viewing angle and can see more of the road in-front of them.
85 MPH speed limit. Long flat straight driving area. 85Mph isn't that crazy of a speed. Especially as I have seen New Englanders go 85mph on roads safe at 55mph.
How does a gun rack create bad driving? A best I would say it may block your field of vision?
Austin is also a
Re: (Score:2)
Austin != Texas. Here, the average car on the streets is not a F-150 (although it is everywhere else in TX), but a VW Jetta, a Mazda 3, Prius, or another compact/subcompact car. You will get the people with the Suburbans, now that gas has gone down in price, but those will be disappearing (and the people buying the Lexus RX SUVs) the minute oil goes over $100 a barrel.
Austin really doesn't have broad streets. As another downside, the city council is in the middle of a project of re-striping most four lan
Re:Question (Score:5, Funny)
Are Texan Drivers worse than Californian Drivers?
I've seen people in Texas driving slow who will pull over to the side to allow faster traffic to go by. This sort of courtesy may just blow the autonomous cars' circuitry after dealing with California drivers.
I expect that the human backup drivers are still from California, so it is unlikely they will be able to take over in that situation either.
Re: (Score:2)
Re: (Score:3)
More likely the driverless car would be the one in front with its bumper being ridden.
In other words, now the self-driving car needs to learn the customs and cultural driving quirks of the local population, otherwise Google's self-driving automaton is at the risk of being declared to be a jackass, for not pulling over for the cars on its bumper to pass.
Re: (Score:2)
Actually, in MOST states, it is illegal to be driving in left lanes with cars behind you. As a law abiding car, it would likely move over like people should.
Human shields (Score:2)
Any bicyclist can tell you the bullying attitude they get from a certain percentage of pickup-truck drivers. Texas is practically the home of the jacked-up pickup.
I guarantee you that if there weren't actual people in those google cars in Texas, they would be getting run off the road (and perhaps run over) with regularity.
Re: (Score:2)
How many drivers does it take to drive a driverless car?
Well, normally it would take one. But this is Google we're talking about here. So with each driver, you have to factor in 20% personal interest leave, nap leave, cafeteria time, sympathetic maternity leave, mandatory Frisbee/unicycle breaks, etc. So you're probably talking 2-3 employees minimum.
Re: (Score:2)
Engine.dll
ExhaustManagement.dll
SteeringBrakes.dll
IRRangeFinder.dll
BackupCamera.dll
CupHolder.dll
Re: (Score:3)
AI-Module_FleeAccidentScene.dll
AI-Module_GetawayCar.dll
AI-Module_MobileProstitutionVan.dll
AI-Module_RamRaiding.dll
BulletProofGlass.dll
HeavilyArmoredBody.dll
InfraredJammer.dll
HoodMountedLaser.dll
BumperAttachedBatteringRam.dll
DriverSideRocketLauncher.dll
WindowTint.dll
StuddedRunflatTires.dll
PerformanceExhaust.dll
OdometerSetback.dll
ColdairIntake.dll
MufflerDelete.dll
RadarDetector.dll
FakeLicensePlate.dll
HiddenRearGunTurret.dll
EMPGenerator.dll
PoliceRadioJammer.dll
SmugglingCompartment
Re: (Score:2)
Re: (Score:2)
Too many. And then you have uninstall and install to upgrade them. :P
What an idea (Score:1)
Re:What an idea (Score:4, Insightful)
Wow, checking if something works outside of California. What an idea!
Wake me up when they test winter driving in upper New Hampshire.
Re: (Score:2)
Wake me up when they test winter driving in upper New Hampshire.
Even summer driving in certain parts of New Hampshire can be a challenge. I would love to see two driverless cars pass each other on certain parts of the Mount Washington auto road [mtwashingtonautoroad.com].
Re: (Score:2)
BINGO. That will be the trick. Although, supposedly, they have made great progress dealing with fog and heavy rain in the Oakland / SF Bay area, which I was skeptical about. Of course, guessing where the curbs and pavement markings are under the snow may need a lot more processing power.
Google is still keeping things very quiet, but from the couple of presentations I've been to, it also still looks like the driverless cars need always on access to the cloud. In Oakland and Austin that's not necessarily
Re: (Score:2)
it also still looks like the driverless cars need always on access to the cloud.
That's no good..... the cloud isn't reliable enough.
Suddenly, you have an outage: all the self-driving cars are stranded and can no longer move, and the world looks like a scene from One Second After [wikipedia.org].
Hopefully they're planning on technology advancing to the point where all the processing power can actually fit in the car, before they would think of finalizing their product.
Re: (Score:2)
I think the real challenge will be like situations the parent described. Mt. Washington would be a excellent test.
I live in the Shenandoah valley. Its mountainous terrain and rural. We have alot of roads that are unmarked in terms of speed, have no center or edge lines, are rather narrow, and in most cases have either no shoulder or dirt or grass the county keeps mowed.
In theory these are 55 zones. In practice people go much faster than that anywhere its flat and strait. The hills offer plenty of blind
Re: (Score:2)
Yeah, that will be an epic fail.
It would be a hilarious test to watch. But an autonomous car in a bad winter snow storm? I don't see that working out very well.
And then you have a car which is either only usable part of the year, or in certain places.
Re: (Score:2)
Don't you watch TV.
Every 80% of the American population live in California, 10% New York City, 3% Chicago, 2% everywhere else.
There is a 1/3 chance that someone is an aspiring actor/actress/comedian. Those who do not live in the city are somehow poor/less civilized. And live in shabby brown homes.
I sugest they stay well clear... (Score:2)
of the book repository.
Re: (Score:2)
Re: (Score:2)
Considering that the book depository that you're thinking of is about 200 miles away, and in Dallas, I think they're safe.....
Yes (I've been there, didn't get to look out the window). I was idly thinking more that Dallas has a history of hostility towards change (Kennedy was jeered, spat on, and abused when he first visited - by people in furs and suits) and is just up the road - and maybe that's part of the reasoning behind the the testing in Texas. i.e. if the reception isn't too hostile in Texas it maybe safe to test and deploy elsewhere.
It was idle speculation that they'd drive north (the I-35?) and test the reception - cautio
Re: (Score:2)
I-35 (Texans don't use "the" with highway names) is particularly fun because it is almost always under construction somewhere. Currently the major construction is between Austin and Waco, and in northeast San Antonio. That is (was) some of the oldest sections (1960s era) still remaining, back when they thought curbs were a good idea on freeways. The Austin to Waco section has long been beyond its capacity, with construction making things worse.
And there was a major incident a few months ago when an over-he
Re: (Score:2)
I-35 (Texans don't use "the" with highway names)
They sure do. Some drop "the" as part of the colloquial vernacular - but most anywhere in Texas I hear directions to drive "up the road/highway"
Roadworks should make an interesting challenge for automated driving.
And there was a major incident a few months ago when an over-height semi truck went under a bridge and pulled out a concrete beam behind it. (The beam was for a new bridge that was still being built, so it wasn't tied in yet.)
We get similar stupid. Trucks jammed under the bridge height warning signs, people suffering head injuries by height warning signs in the on-ramps to multi-story parking because a passenger gets out and tries to swing the warning out of the way so an over height vehicle can enter. We also get stupi
Re: (Score:2)
Re: (Score:2)
So far, all i've heard is here but you've got a point about Texas being resistant to change. Ironically, Austin (the people who live there, not the government) seems to be one of the more flexible parts of the state, that may be why they chose there first.
I don't want make it sound like Texas is the most resistant (or the most vehement in their resistance) of states - I can think of others (and I reside in Australia). East Texas is quite different from West Texas, landscape, weather, and the nature of people - but the differences in attitudes is not uniformly so. From an outsiders perspective people in Dallas (many that I meet weren't born there) don't seem much different from Austin - but the government sure is. Politics in Dallas doesn't seem to have chang
Texas? (Score:4, Funny)
It shouldn't take long for some of the inhabitants to consider this to be the tip of the Obama Administration spear to take over Texas so they can remove their guns, impose environmental regulations, force money to be spent on education. And this right after the Jade Helm 15 exercises. They are probably Islamic driverless cars.
Re: (Score:2)
Nah. It doesn't say anything about pickup trucks, momma, dogs, or the rain.
David Allen Coe [google.com]
Re: (Score:3)
Re: (Score:1)
Comment removed (Score:5, Informative)
Re: (Score:3)
I honestly don't know how it's coded to behave in such an incident
if (victim.LicensePlate==DigiShaman)
{
perform_hit_and_run();
}
Re: (Score:2)
Re: (Score:3)
You know this is in Austin, right?
It's more likely that they'll soon have the car running entirely on vegan fuel, maybe with a purple hair dye job.
Yes, people hear Texas and think guns and pickup trucks. But Austin is basically what it would be like if California was in Texas.
Re: (Score:2)
Re: (Score:2)
Based on his comments, he knows nothing about Texas other than the sterotypes that Hollywood pushes forward.
You don't expect people like him to actually not believe everything they see on TV - other than Fox News, of course? I'm sure he has been warned (and even seen 15 second out-of-context clips to reienforce his belief) on the evils of Fox News.
Re: (Score:2)
It shouldn't take long for some of the inhabitants to consider this to be the tip of the Obama Administration spear to take over Texas so they can remove their guns, impose environmental regulations, force money to be spent on education. And this right after the Jade Helm 15 exercises. They are probably Islamic driverless cars.
Google? They're right on time: gathering intelligence for the upcoming JADE HELM 15 invasion.
Re: (Score:2)
Austin, that had already happened.
Two drivers (Score:2)
Ah, but the real question: (Score:2)
Are they dealerless as well as driverless?
It would be ironic if they could be driven by a computer, but have to be sold by a human in Texas.
This run at driverless cars will fail (Score:5, Interesting)
OK here's the thing with this generation of driverless cars- their motion is governed by neural nets. I am going to assume that everyone here is familiar with this programming paradigm. If not, the Wikipedia entry on it is adequate.
While in the end NN are just another form of Turning machine, currently no one can divine the algorithm of a trained neural net well enough to express it in IF THEN ELSE WHILE form.
That means given a trained NN which is 100% correct 100% of the time , no could write an imperative or procedural (broadly speaking) program which captured the logic (IF THEN ELSE) the neural net is using (defacto using, NN don't have IF THEN ELSE logic except those implicitly embedded in their activation rules) to solve the problem.
That means the algorithm the NN has arrived at is not open to analytical inspection and confirmation, except very indirectly.
This is OK for wide variety of predictive tasks in which human life does not hang in the balance. In medicine, the diagnostic results from NN and even Good Old Fashioned AI expert systems are reality--checked by human doctors.
Neural nets ALWAYS run the risk of coming to the right conclusion for the wrong reason enough of the time to fool humans into thinking it "understands" the problem domain in a way that is analogous to a human. A NN so trained will fool or lull human observers into a false sense of security until that BIG ACCIDENT happens then a post mortum reveals the shocking truth about what the NN was focusing in on to make it decisions.
The Big Idea behind NN is that, through a combination of evolutionary forces and billions of iterations the NN will learn using the same Hebbian activation princples the brain appears (now) to use and that with enough training, the exceptional cases that I am describing will be found and rooted out.
But even in nature, this doesn't happen reliably. Take for example the Australian Jewel Beetle. Over perhaps millions of years, it has of course evolved a robust way to recognize desirable mates and procreate. That is as basic an evolutionary task as you can imagine- it has to work or the species is doomed.
However, the male's algorithm for mating is not as robust as you might imagine. It seems that what males rely on to select a mate is a very, very limited set of perceptual cues. As it turns out, it is looking for big glossy brown curved things. When it sights one, it alights and starts humping away.
Well, Austrailian beer bottles fit this description *and fit it better than the female of the species*. People toss empty beer bottles in the outback and the result is the male beetles prefer the beer bottles to such a degree that the beetles were going to go extinct. Austrailia had to pass a law to change the appearance of its beer bottles.
http://blogs.scientificamerica... [scientificamerican.com]
This is a cautionary tale to those who think evolutionary forces produce only *robust* algorithms. What evolution actually produces is *good enough so far* algorithms. What well trained NN produce are similarly good enough algorithms. In both cases we have to do science to try to get at what it is they are relying on- what features they are *really* trained on. And we don't know there's a problem until tragedy happens and we don't know how ridiculous the problem is until we do science.
This is different from procedural programming which, the Halting Problem notwithstanding, CAN be analytically examined for correctness. Procedural type programming plus sensors is what runs water stations, trains, planes etc. The military does use NN to try to recognize things but it has humans making the final decision and when the missle gets launched, it's not left to a NN to decide where to finally land.
Moreover, self driving cars under the control of a NN can and will be attacked by the usual miz of 14 y/o kids, pranksters, criminals and terrorists
Re: (Score:1)
If Google are prepared to insure their own cars, then frankly I don't see any real obstacle to widespread adoption.
The statistics so far suggest that the cars are already safe enough (better than humans), and will become safer.
Re:This run at driverless cars will fail (Score:5, Informative)
Hi, I'm a developer who works on intelligent vehicles.
We don't use neural nets, for exactly the flaws you point out.
Re: (Score:2)
Cool, and what's the procedure you use to tell the difference between a plastic bag (which you can run over) and a small child (dead stop)?
Re: (Score:2)
Thing is, the systems that are already on the road are also controlled by neural nets. Really crappy ones, with slow reaction times, a very limited sensor set (it has barely 110 degrees of vision), and is incredibly prone to impairing even those limited abilities. Yes, it's got a few advantages, with an almost preternaturally potent visual recognition system, but even so it's responsible for 10 million accidents per year, with tens of thousands of fatalities.
Humans just aren't very good drivers. Automated s
Re: (Score:2)
Your point is good but I think it ignores the fact that we have an entire system of law built up to assign culpability to humans, in fact, that is mostly what the law is about- holding humans accountable "as if" they could be held accountable, as if their actions were not compelled , which I also think is in very many cases a fiction.
So we have this body of law and historical precendent worked out in ultra-fine detail and a very very deep seated belief that people ARE responsible for their inattentiveness a
Re: (Score:2)
As long as the total liability is decreased, then it's a solvable problem. We already have mandatory insurance in a lot of places. We could piggyback off that. It would involve some legislating and contracting and other paper-shuffling, but it doesn't seem impossible to just treat it as part of your insurance.
Heck, the insurance companies might even offer you a discount for turning your car over to a superior mechanical driver.
People may want to go after Google's deep pockets, and that's up to lawyers to fi
Re: (Score:2)
I think it's a solvable problem.
As long as the manufacturer has a sufficient income stream and a way of making sure that cars with known flaws are fixed there is no reason they couldn't cover the liability for all their cars and they of course have the option of taking out an insurance policy against that eventuality. The key will be ensuring that revenue stream. The nightmare situation for a manufacturer is being held responsible for a product they no longer make any income from or have any control over.
Fo
Re: (Score:1)
I took Sebastian Thrun's excellent class at Udacity concerning programming AI for self-driving cars. There were no neural networks involved. Basically, it is about using sensor data, known maps, and control of steering and velocity, to stay on the road while maintaining safe distances from other objects.
Re: (Score:2)
We don't need to be 100% safe, safer than human drivers is enough.
Plus, if there are indeed neural nets, they are not necessarily relied upon in life-or-death situation. In fact, they most likely aren't.
For example, the prankster tricking the car into thinking the bridge abutment is a road may fool some advanced AI until another, much simpler piece of code tied to a proximity sensor triggers an emergency maneuver. Rough ride but you are safe.
And, I wouldn't call these people pranksters. Murderers would be a
Re: (Score:2)
Boston (Score:3)
You want to really give those cars a test? Boston. Or New York City, if you must. But Boston would be better. Lanes that disappear without warning, roads that are blocked, randomly, by "repairs", signs warning you of work zones (and double fines) which then turn out to be devoid of any workers (indeed, any sign whatsoever of work, except for a few orange plastic barrels), lack of any signs when you need them most...I could go on -- but I bet Google's cars couldn't!
And then, Boston drivers...
don't need no steenkin' computers to drive Boston (Score:2)
Re: (Score:2)
You forgot Los Angeles (L.A.) for its #1 traffic!
Drunk pedestrians (Score:2)
That is great news, the self-driving cars will now be able to test themselves against more drunk & rowdy pedestrians in Austin!
Driverless* (Score:3)
*When the two humans driving the car feel safe enough to let the computer take over, which according to Google's own data [googleusercontent.com] is about half the time.
So sick of hearing about this driverless car bullshit, when it isn't anywhere near that.
snow (and manholes) (Score:2)
So, test it in Alaska, in winter.
And on a road with open manholes... :) Did they solve that yet?