Slashdot Mirror
The Documents From Google's First DMV Test In Nevada
An anonymous reader writes "IEEE Spectrum contributor Mark Harris obtained a copy of the DMV test Google's autonomous car passed in Nevada in 2012 and associated documents. What has not been revealed until now, is that Google chose the test route; that it set limits on the road and weather conditions that the vehicle could encounter; and that its engineers had to take control of the car twice during the drive.
I'm sure the editors are shocked and amazed that "Google chose the test route; that it set limits on the road and weather conditions that the vehicle could encounter; and that its engineers had to take control of the car twice during the drive."
Don't waste your vote! Vote for whoever you want, unless you live in a swing state it won't matter anyways
All of the recent articles about autonomous cars seem to be trying to make people think they're terrible will never work and are a disaster waiting to happen.
All of these tests and such aren't being done so they can release an autonomous car tomorrow, its an ongoing process and will take time. I don't blame google for not wanting to publish all the details about it, its a research project and the media seems to have an agenda to make autonomous cars into the boogeyman
I'm really not that suprised. The sheer amount of decisions that we make while driving (which is a pretty complex and chaotic system) is pretty staggering. Not to mention you have local/state/national laws that vary - just to make it a bit more challenging. I really don't forsee driverless cars becoming a reality until local communication between all cars can resolve a lot of problems.
The conditions are controlled and constant in order to test the system's response, to ready the system for variable conditions with solid behavior data.
From TFA: 'In communication with the Nevada DMV before the test, Google said its policy was to prohibit autonomous operation at railroad crossings that lack signals and for human drivers to take over. It also noted: “[Roundabouts are] particularly challenging, where many drivers don’t know the proper rules in the first place.” In an e-mail to colleagues at the DMV, Breslow wrote, “We can’t fail an applicant for not being able to navigate a traffic circle if they say that there [sic] vehicle can’t yet do it.” '
So the two times that the Google engineer took over were for the two things that Google said they felt it was unsafe for the car to handle - a railroad track without a signal, and a roundabout.
Occasionally living proof of the Ballmer peak.
Sure sounds like a research experiment vs engineering a solution.
Like any simulation, you can tweak the knobs to get the answer you want.
Interestingly, this quote from the bottom of the page says it all "Any sufficiently advanced technology is indistinguishable from a rigged demo."
Let me know when these things can drive themselves in the rain and snow. Until then, yawn.
Brave Sir Robin ran away. ("No!") Bravely ran away away. ("I didn't!")
We cant fail an applicant for not being able to navigate a traffic circle if they say that there [sic] vehicle cant yet do it.
Bullshit. in this case the vehicle is the driver. the car itself is entirely capable of turning and entering a traffic circle but in this case the driver, the autonomous car, wont do it. the excuse from google "most people dont know how to navigate a traffic circle" is also some supreme bullshit. most drivers in practice never signal their lane change, drive much faster than the speed limit, and overtake vehicles improperly but those inherent dangers didnt mean the autonymous car couldnt negotiate a highway, it simply meant they were ignored because google got to cherrypick a desolate strip of highway. this wasnt a driving test, it was a publicity stunt.
When faced with a partially blocked-off road, the car switched between autonomous and manual modes and then braked to a halt. Wojcik also recorded that the car needed driver assistance with some turns, although she did not note the circumstances.
Road construction is a fact of life. Are we to seriously believe Nevada thought it was entirely acceptable for the vehicle to come to a complete standstill in construction traffic? and problems turning? thats kind of a fundamental of motor vehicle operation that should you mess it up will generally not see you pass the exam. the fact that this drivers exam was only made available after a FOIA request is also a red flag that nevada isnt taking this seriously. should this car be allowed as its been passed onto Nevada roads, and in fact be responsible for a major collision, this document will come back to haunt quite a number of politicians.
Good people go to bed earlier.
"engineers had to take control of the car twice during the drive." You mean like many of our parents did reflexively when we were learning how to drive? Not to mention the dent in the passenger side floor where the extra brake pedal wasn't.
"Win treats sysadmins better than users. Mac treats users better than sysadmins. Linux treats everyone like sysadmins."
This shows that the claims of Google have been overhyped, but while I have no reason to trust Google, I have to wonder if the fault is Google or the people who pretend to be journalists.
Maybe if they say the car won't handle it and the car correctly recognizes the situation and shifts into manual mode, that's a pass. Makes sense.
Peter predicted that you would "deliberately forget" creation 2000 years ago...
I guess so.
Car: "Warning: 500-foot cliff detected. Switching to manual mode in 3...2...1..."
Human: "AAAAaaaaaaaaaaaaaaaa......"
TEST PASSED!
[Napoleon Dynamite straps himself into the time machine]
Kip: It's a time machine, Napoleon. We bought it online.
Napoleon Dynamite: Yeah, right.
Kip: It works, Napoleon. You don't even know.
Napoleon Dynamite: Have you guys tried it yet?
Kip: [reluctantly] No.
Kip: So are you ready?
Napoleon Dynamite: Yeah, hold on... I forgot to put in the crystals.
Napoleon Dynamite: [using time machine] Ow! Ow! Ow! Kill the pow... It kills! My pack! Ow! Turn it off! Turn it off, Kip!
[Kip pulls the electrical cord out, and Napoleon yanks off the headband]
Napoleon Dynamite: It's a piece of crap it doesn't work!
Uncle Rico: Well, I could've told you that.
[Uncle Rico is standing in the hall, looking wounded and disappointed]
Big Ben! Parliament!!!
âoeWe canâ(TM)t fail an applicant for not being able to navigate a traffic circle if they say that there [sic] vehicle canâ(TM)t yet do it.â
They had no trouble failing me for not being able to navigate certain traffic situations correctly on my first drivers test. Had I said, well "I can't do that yet" that would not have resulted in a pass.
So the car failed. Let it fail. Its no big deal that it failed. Constructing an illegitmate pass for headlines just stirs pointless controversy.
This test shows that there's lots of good progress, but that's it.
As for driverless cars actually being ready ? No where near close.
Requiring the driver to be ready to jump in while in action is absurd to the point of not even considering it.
Stopping the car, and handing the controls over is still going to lead to tons of problems. Cars stopping at railway crossings and round abouts and then just sitting there jamming traffic... because the driver fell asleep 30 minutes ago. (And why SHOULDN'T the driver fall asleep -- he's tired, bored, and not doing anything... what do you expect will happen)
Until self driving cars reach the long term goal of being responsible for driving in basically anything a human is currently expected to cope with they can't rise to being more than a novelty, or some limited highway auto-pilot cruise control system.
Because even if it CAN usually handle the daily commute, if it can't handle it ALL THE TIME its a bad idea.
Today when there's snow in a city that doesn't get snow that often its a mess. And that's with mostly drivers who drive every day, know the route they are going like the back of their hand, know where the tricky / problematic spots are etc, and know know how their cars handle at least in normal conditions. And its a mess.
Now, lets substitute that with a city full of drivers who only drive 3 - 4 times a year, are completely out of practice, have no real experience with their car, and only have a general imprecise sense of the route they need to take -- and lets do that on a day the self driving cars collectively decided they can't handle the conditions.
That would be like me driving my grandmother around every day everywhere she needs to go, and then when the weather is at its worst... "Hey grams, yeah, I know you've only driven this car a handful of times yourself in the last 5 years, but your license is still valid, so how about you take the wheel today?". And doing that accross an entire city.
Yes, that will work out well.
Commercial aviation is now safer than it ever was in the past.
It should also be noted that automated flying is generally done once the plane gets to about 10,000 ft / 3000m. Take-offfs and landings (excluding ILS Cat II+) are generally much more manual. So is taxing at the airport.
I would say the rough equivalent would be cruising on the highway: let the auto-drive handle cruising and stop-and-go traffic in a controlled environment, and give back manual override as the highway exit approaches and the car is on the more chaotic non-highway roads.
Cruising is mostly taken care of, and Audi has announced the stop-and-go part:
http://www.bloomberg.com/news/2014-09-09/audi-plans-to-take-over-the-wheel-in-stop-and-go-driving.html
And if we limited the exam to questions i knew the answers to, i would always pass.
I am curious as to how the license works. Is it for a specific revision of software, or does Google have free reign to make large changes to the software without re-testing?
It seems to me that as more of these vehicles hit the road, even in test mode, there should be tight control over how big of a change can be Beta tested on live streets without a re-qualification.
Everyone criticizing them seems to be missing the important part: We need self-driving cars so we can have self-driving flying cars. I don't trust most people to handle a flying car but self-driving actually becomes easier when crossroads can be handled on different positions on the Z axis. Don't ruin it, jerks!
The biggest failing autonomous cars have is assuming they are *not* surrounded by malicious, aggressive, incompetent, brain-dead animals.
The thing to remember about self-driving cars is that they're nearly on-par with humans now, were crap ten years ago and not even that twenty years ago, whereas human drivers are on-par with human drivers ten years ago, and exactly the same twenty years ago. And thirty. And forty. Technology is improving like crazy. Humans aren't.
People keep saying to me "What if there's a bug that causes lots of crashes?" Well, we'll fix it. It might take a few days, or a few years, but we'll fix it. Right now, human drivers have a collection of bugs causing an average of 110 traffic fatalities per day in the US. We can't fix those bugs--drunkenness, fatigue, eating while driving, using your phone while driving...we tell people to shape up, but they just don't.
A. A railroad crossing without signals
B. A roundabout
C. Construction work
D. "Some specific turns"
As it approached the railroad crossing the Google Car coasted almost to a halt, at ~1% of full power. This was due to an excess 'poisoning' of Xenon-135, a persistent condition that was generally understood but specifics directly relating to operational safety at low power had not been addressed. At this point conditions for the driving test were inherently unfavorable and dangerous. Tuptunov sensed this, but he lacked the knowledge, vocabulary and resolve to communicate this to Dyatlov -- who ordered the car to be driven manually over the crossing.
This led them into the roundabout, where a single path to the destination exists but the Google Car was not configured to confidently know how and when to exit. The automated systems drove the car in circles for several minutes in low gear at high RPM. The car was still in a state of equilibrium and would eventually have allowed the excess Xenon to absorb neutrons and decay to Xenon-136, which has a much smaller cross-section. But again Dyatlov was impatient for the test to complete and he was getting dizzy, so he ordered to withdraw all but six of the control rods and manually lurch the car into the turnoff.
At this point the car was screaming at full RPM in low gear as it approached the Construction Zone, lurching and swaying. The operators knew they were in some sort of trouble, but the Google Car jerked forward automatically until it spotted the red cones and barricades. It disengaged to manual control and Tuptunov slapped the vehicle into its lowest possible gear. Under normal conditions this engine-assisted braking procedure was the best possible course of action. But the pistons and rods were tipped with graphite which causes a temporary neutron flux when inserted, which escalated power and deformed the rods.
At this point things in general took "some specific turns" for the worse.
<blink>down the rabbit hole</blink>