How Autonomous Cars' Safety Features Clash With Normal Driving
An anonymous reader writes: Google's autonomous cars have a very good safety record so far — the accidents they've been involved in weren't the software's fault. But that doesn't mean the cars are blending seamlessly into traffic. A NY Times article explains how doing the safest thing sometimes means doing something entirely unexpected to real, human drivers — which itself can lead to dangerous situations. "One Google car, in a test in 2009, couldn't get through a four-way stop because its sensors kept waiting for other (human) drivers to stop completely and let it go. The human drivers kept inching forward, looking for the advantage — paralyzing Google's robot." There are also situations in which the software's behavior may be so incomprehensible to human passengers that they end up turning it off. "In one maneuver, it swerved sharply in a residential neighborhood to avoid a car that was poorly parked, so much so that the Google sensors couldn't tell if it might pull into traffic."
Ban human drivers.
"One Google car, in a test in 2009,..."
One would think that in 6 years some improvements would have been made. Do we have a more current example?
09 F9 11 02 9D 74 E3 5B - D8 41 56 C5 63 56 88 C0 45 5F E1 04 22 CA 29 C4 93 3F 95 05 2B 79 2A B2
Let them come up with the self-driving algorithms..
Millions of people on the road today deserve to have their license taken from them because they can't follow simple rules like signaling, not parking halfway out into the street and leaving enough room to brake in case the car in front of you brakes.
The article summary isn't very good. If the software is programmed in a way that causes a car to behave in a way that's dangerous, it IS the software's fault. This is different from a hardware failure or driver error. It is the software's fault. The issue is that the car's programming is functioning as the designers intended it to, but this leads to dangerous behaviors and unexpected consequences. That's different from the software behaving in a manner inconsistent with what its designers intended it to do, which is what the submitter was trying to say wasn't the cause. However, both of these are still the fault of the software.
From what I have read from people who have actually interacted with a autonomous car. They are very pokey, slow, and tend to pause trying to figure out what to do.
In fact many times its the required human driver who has to intervene in order to help the car out of a jam. I think the more we try and mix these auto driven vehicles with human one's the more we will experience the growing pains of this technology.
“They have to learn to be aggressive in the right amount, and the right amount depends on the culture.”
Very true, When holidaying in Texas I quickly found out that stopping for a red light that had just turned would upset drivers behind me. The lights had a much longer amber time, so a whole lot of people who would have had to brake for the lights in the UK would go through
It's not normal to disobey simple rules or park like a retard.
They are confused by BAD driving. People in general really really suck at driving and a computer will have problems with that.
Do not look at laser with remaining good eye.
If the programming makes it jerk the steering away from a stationary hazard rather than, say, detect it earlier and slow down as it approaches, then it's not suitably programmed for coexistence with unexpected stationary hazards (Not even anything to do with human presence! What if that was a cardboard box and it swerved heavily in case that box "pulled out"?).
If it can't make it's way through a junction where the drivers are following the rules, that's bad programming. If it can't make it's way through a junction where other drivers don't come to a complete halt for it, it's not fit to be on the road with other drivers.
If you want a car to co-exist on the road, it has to be treated as a learner driver. If a learner driver swerved at a non-hazard, they would fail. If a learner driver refused to make progress at a junction because the masses didn't open up before it, they would fail. So should an automated car.
Unless - and this is important - you are saying that automated cars should only operate on automated roads where such hazards should never be possible and they are deliberately NOT programmed to take account of such things. Which, in itself, is expensive (separate roads with separate rules with no human drivers), stupid (that's otherwise known as a "train line", and because they can't do anything about it it will hurt more when it does happen), and dangerous (because what happens if a cardboard box blows over the automated road? etc.).
Program to take account of these things, or don't plan on driving on the road. The safety record is exemplary but equally there are only a handful of them and the eyes of the world are on them, and there are still humans behind the wheel, and even by miles travelled each one is probably dwarved by a single long-distance driver over the course of a year - and it's not hard to find a long-distance driver who's not had an accident for years.
If you're going to be on the roads, then you need to be able to take account of all these things, the same as any learner driver. Sure, you didn't hurt anyone by swerving or not pulling out, but equally - in the wording of my first driving test failure - you have "failed to make adequate progress" while driving.
A car sitting on a driveway would have an even better safety record but, in real life, it's still bog-useless compared to a human. Similarly for any automated vehicle that just stops at a junction because it can't pull out, or swerves out of the way of a non-hazard (and potentially weighs up collision with non-hazard vs collision with small child and gets it wrong).
What about speeding? Even more so on under posted highways / interstates / toll roads?
Useing the center of the road as an extended trun lane? even when not marked as one?
rolling stops when no other cars are in the way?
This. All the studies that I've seen boasting about the enormous time advantages of self-driving cars ignore the fact that most human drivers tend to cruise from 5 to 15 MPH over the posted speed limit on many interstates and highways. I can't imagine a self-driving car being designed so as to operate above the posted speed limit in self-driving mode. Unless a second set of roads or a second set of rules is created for autonomous vehicles, you're going to have a difficult time convincing people of the advantage of being slower than anyone else on your morning commute.
Isn't the point of the road tests to get through the growing pains of this technology? I think as we mix more autonomous vehicles with human drivers people will eventually learn to drive more safely themselves.
"the accidents they've been involved in weren't the software's fault"
That's *GOOGLES* claim, and to back it up they released ONE incident analysis, which represented their cars interpreted view of the world. I want to see *ALL* the footage released and ALL the interpretation data released.
You could get rid of 4 way stops which are retarded anyway.
The article summary isn't very good. If the software is programmed in a way that causes a car to behave in a way that's dangerous, it IS the software's fault.
No, it is the programmer's fault. Software is an amoral set of machine instructions written by a human. Saying it is the software's fault is like blaming a press for cutting off someone's hand. The actual fault is either user error or faulty machine design. The machine is just doing what it was told so blaming software is misplaced. The fact that the problem of autonomous driving has a lot of difficult and dangerous corner cases is irrelevant.
Software is just a set of instructions given by a human so if the instructions are wrong it is the fault of the person who gave the instructions to the machine. If the car behaves in a way that is dangerous then it is the fault of the person/company/entity that wrote the software controlling that behavior. The programmer is just as much at fault as a driver who made a mistake. The difference is that the programmer probably isn't sitting in the car at the time but just like the driver he/she was the one in charge of the driving of the car. It's really easy to forget that a human is instructing the machine - the difference is that the instructions are time shifted.
I think in time autonomous cars could prove to be safer than many/most human driven ones. There are a LOT of really bad drivers on the road and the tests to qualify for a license are pretty much a joke and there are almost no requirements to re-qualify which is kind of nuts. But the liability for bad or inadequate software should fall squarely on the party that wrote said software.
Actually, autonomous cars are programmed to exceed the speed limit by up to 10 mph. This is done because Google deems it safer than driving at the speed limit and being slower than the other cars on the road.
http://gizmodo.com/googles-autonomous-car-is-programmed-to-speed-because-i-1624025227
http://www.bbc.com/news/technology-28851996
If all the cars were autonomous the morning commute times could be cut in 1/2 or 1/3rd without changing the speed limit since rush hour style rubber band stop and go traffic would be a thing of the past.
The truth is most people suck at driving. Genuinely good driving takes effort and concentration and most people however skilled or well-intentioned have got other stuff on their minds when they get behind the wheel. There's no hope of ever getting people actually to drive according to the rules so the only way is take the human driver out of the loop completely and the sooner the better.
"Don't belong. Never join. Think for yourself. Peace." V.Stone, Microsoft Corporation
Not necessarily. Ask yourself, why do people speed; I suspect that in many cases it's because the driver thinks of their time driving as time wasted. If the occupant no longer has to drive and can instead focus exclusively on other things, it's a real game changer.
Even if the car would be programmed to follow the rules exactly, how much time would you actually lose on your daily commute? Really? Especially when weighed against the fact that you can spend the drive to work reading / working / making calls or whatever.
If construction was anything like programming, an incorrectly fitted lock would bring down the entire building...
So is Google going to pay my speeding ticket when a cop pulls over my autonomous automobile for speeding?
Demented But Determined.
with ABS, airbags and selectable 4WD. Window cranks, inert key and levers on the seat. Cheaper, less to break, and I think I can still manage.
"Win treats sysadmins better than users. Mac treats users better than sysadmins. Linux treats everyone like sysadmins."
So is Google going to pay my speeding ticket when a cop pulls over my autonomous automobile for speeding?
Almost certainly. Though they will bring in several well-respected highway safety engineers to testify that following the flow of traffic is significantly safer than following the posted speed limit. Enough jurisdictions will lose money arguing these cases that there won't be money to be made by writing the tickets. Absent both the financial and safety benefits the police will stop issuing the citations.
And deer. And snow too. Also road construction.
Most certainly not. Eventually there will be a class action suit, though, and a firmware upgrade will allow you to force the car to strictly obey the limits. That will be about the same time that there is saturation of self-driving cars and the sheer number of them that are keeping speed to within 0.01% of the posted limit will ensure that nobody can speed. Municipalities will then all complain of the lost revenue.
And even if morning commute times were longer, who cares, as it wouldn't be commute time. You would start working the second you hit the car. Who cares if it takes two hours each way if you have your computer open doing work.
Unsafe is usually interpreted to mean using an acceleration that may not be available to all vehicles.
No it isn't. Unsafe is operation of a vehicle in a manner that violates duty of care. You can safely accelerate faster in a Corvette in many circumstances than the maximum acceleration of a Nissan Versa. Whether it is unsafe will depend on factors including road conditions, visibility, nearby traffic, nearby pedestrians, driver skill, vehicle capability, etc. There is no requirement to only accelerate as slowly as the slowest vehicle.
Obviously a sports car on special tyres can stop a lot faster than an old banger or a truck, and and therefore you should never use maximum deceleration (emergency stop).
You use emergency stops in an emergency. It is the responsibility of drivers behind you to maintain assured clear distance part of which is evaluating the likely possible deceleration of the vehicle ahead of you.
Whether the driver behind you is paying attention is not your problem.
It becomes your problem when they crash into the back of your car. I've had that happen several times, including several when I was stopped. Every time the other driver was ticketed for failure to maintain assured clear distance. Drivers around you are always your problem if you are driving defensively.
Anybody not paying attention at a traffic light should not be on the road.
And can you honestly tell us that you have never once gotten distracted when the light changed? If you claim that you haven't you either drive very very little or you are lying. It happens to every driver now and then.
https://www.youtube.com/watch?v=Fbpo6rOHCZU
Would You Buy a Car That’s Programmed to Kill You?
And then I will work in the back while my personal tour bus takes me cross country.
if you get there at same time as someone coming from opposite direction (i.e. no one on right to which to yield)
And that's a ticket for each one of those offenses [if caught]. I've been ticketed in the last 5 years for each and every one of those. Annoying? Yes. Unavoidable? Don't use a Google car. :)
No points left, but this and the parent, too.
I have visited Taiwan a few times, and each time I am scared when I ride in a car. Drivers swerve out of their lanes or pull out suddenly into traffic. Scooters weave all over the place. Crossing busy intersections is very different from American modes of driving. If I were to drive in Taiwan, I'm sure I would get into a daily accident. However, the locals seem to be able to manage what I see as chaos because they anticipate these driving behaviors and compensate by driving more slowly and continually looking out for actions that are not necessarily base on written law.
On the other hand, a Taiwanese driver may have difficulty driving in the US due to higher speeds and more than occasional intransigence at protecting one's legal right of way even at the cost of safety.
The idea I'm trying to convey is that the driving system seems to work if all drivers follow the same philosophy, and anticipation of other drivers' behavior matches reality. This is one of the problems with self-driving cars. They may be "perfect" at following the law and "common sense", but if their "common sense" conflicts with that of surrounding drivers, there's a potential problem, even if others are not technically following the law. Take for example the car driving at the speed limit on a highway where everyone else is driving 10-15 mph faster. The slow car is the only one technically following the law, but the speed differential from that one car potentially causes problems.
About those sensors.... my late wife once had an incident: the light turned green, but for some reason she waited... and just then a semi ran the light. Self-driving car - 0, semi-1.
Or the times where you're stuck behind an idiot doing 5 or 10 mph below the speed limit, in the passing lane; sometimes, if you need to get somewhere on time (say, a job interview, or a reservation), you need to time it and take a chance to get around them.
I have grave doubts that the self-driving car sensors do the following:
1. Look *under* the car ahead, which is large, or has ultra-dark sunscreening, to see if the
car ahead of *them* has just hit their brakes.
2. Do the equivalent of turn your head really rapidly, to see if the car on your left or right, that
was a good ways back has just sped up to keep you from getting in front of them.
3. Look at the car on your left or right, and predict that they're going to change lanes,
with no blinker, into a space that's unsafe in front of you (and then hit the brakes)
Oh, and and have auto-updates, with no fees, that are secure (HAH!), for as long as the vehicle is licensed (says the guy who keeps his vehicles 10 years or more). I mean, how many of *you* reading this have taken your vehicle in for *all* the recall fixes, esp. if you bought it used?
mark
If all the cars were autonomous the morning commute times could be cut in 1/2 or 1/3rd without changing the speed limit since rush hour style rubber band stop and go traffic would be a thing of the past.
Rubber banding cannot stop.There still has to be a delay between a car in front to start moving and the car behind to start moving, in order to maintain safe separation distance. 10 feet is fine at a stop, but as soon as you are going 1 mph, 10 feet is no longer safe. So the car ahead needs to travel 5 or 10 feet before the one behind can even start moving. The opposite is true when stopping for a light, a stop sign or a road obstruction.
Commute times would be much longer because the autonomous cars would allow a safe distance between each other. This would require either fewer cars on the road, or a slower speed so that the existing number of cars on the road are all maintaining safe distance for the speed.
If you are not allowed to question your government then the government has answered your question.
And how can you know what they'll do when they're not trying to cut in front of you too close or turn in front of you or tailgate you or run redlights or the other things you expect others to do?!?!?
I expect we'll do away with speed limits altogether. They will probably be replaced with an acceptable speed range. For example, where the speed limit is currently 55 the acceptable speed might be 50-60 mph. Once you integrate vehicle-to-vehicle communications this speed can change with conditions. The speed range on an icy interstate might be 40-50. That same section of highway might go to 70-80 when it's sunny and dry. http://www.autonews.com/articl...
I do not block ads. I do block third party scripts.
About 3 months ago I saw a Google car do something bad for the first time. The Google car had a red light and was attempting to turn right on red, the cross traffic light was signaling a protected left and an SUV was trying to make a U-turn onto the road the Google car was heading. The Google car kept starting and stopping 3-4 times before I assume the human intervened and put on the brakes. It seemed like the car was repeatedly trying to make its right but jerking to a halt as it noticed the SUV. The road was a 3-lane road, I believe the SUV made its U-turn within 2 lanes but can't recall clearly, it may have taken 2.5 lanes which maybe was the cause of the problem.
The evidence suggests this is not true. The reaction time of autonomous cars is much faster and they're rarely mucking around texting and fail to notice the car in front has stopped moving. Nor do they over break when the car in front of them slows down to allow some car to merge. This is where most congestion occurs. Humans over reacting when two major roadways converge, say the interchange of two interstates. Autonomous cars will be substantially better at folding those two streams together.
So yes, pedantically rubber banding will still occur but nowhere near as badly as in human driven vehicles.
Since the outside lane is ONLY FOR PASSING. Those four-six-eight or whatever lane highways are wasting 3-9 lanes of space for NOTHING.
Tell me, if your speed limit is 75mph or whatever, and you're going 75mph in the outside lane, what on earth is the problem here?
And NOBODY in ANY lane *especially* the outside lane should EVER go over the speed limit. If you can't pass in time at the speed limit or less, then you can't pass.
:)
sleeping on your morning commute is pretty awesome.
or you know, slashdotting.
Living in Austin: It takes my gf 20 minutes to get to work in no traffic, and up to 30-40 in rush hour. When I'm a passenger in a car I can't read books or look at cellphone map/screen for more than 5s, it makes me carsick. I'm definitely not expecting to be working as soon as my car leaves the garage.
This reminds me of the problems we've encountered (and keep experiencing) in natural language processing. In fact, in NLP the error margin is fairly loose - 85% accuracy is considered good. Not so for driverless cars.
Having had a close look at the tech behind driverless cars and controlled conditions in which they are tested, I am convinced that driverless cars are the _most unlikely_ outcome of all this research and testing. What the tech could be used for - eventually - is something entirely different - autonomous robots that could move among humans.
I've always been of the mind that a lot of roadway politeness is more because people don't know what to do, so they just let someone else make the decision.
A slow commute isn't such an issue if you can spend it relaxing or working instead of driving... and even speeding by 15 MPH only saves a few minutes on a commute.
This space intentionally left blank
So is Google going to pay my speeding ticket when a cop pulls over my autonomous automobile for speeding?
Almost certainly. Though they will bring in several well-respected highway safety engineers to testify that following the flow of traffic is significantly safer than following the posted speed limit. Enough jurisdictions will lose money arguing these cases that there won't be money to be made by writing the tickets. Absent both the financial and safety benefits the police will stop issuing the citations.
That would solve the problem. I'm quite sure that the judge down in Ludowici, Ga would throw out any ticket if a well-respected highway engineer was brought in from, say, California, to tell them that speeding was OK if everyone else was doing it.
In some places in Europe, you mean. I've seen this in France, but never in Holland, Belgium, Britain, Germany,
FTA - "Despite the awkward encounter, the cyclist didn’t leave with a negative impression of self-driving cars."
Imagine that, somebody driving a fixie who understands testing! That "The two guys inside were laughing and punching stuff into a laptop." means that, yes, they have a sense of humor and the absurd, but are punching stuff into a laptop, presumably to log additional information in for that encounter.
With this incident added to the list, the programmers can now develop a module to handle this behavior appropriately.
I get irked when a problem is encountered in a test and used to imply that the whole idea is stupid. This article isn't too bad, demonstrates WHY they're still doing testing.
I don't read AC A human right
The one making the turn yields if you both arrive at the same time. Just the same as if you both didn't need to stop.
"Lack of speed can be overcome. In the worst case by patience." --Znork
People who get motion sickness would have trouble taking advantage of commute time, so it still matters. There would probably also be workplaces that don't recognize the time spent working while commuting, so if there isn't something you actually want to be doing that you can do in your car, it still matters.
It normally amounts to political. People drive what they feel is safe, normally speaking.
More in depth reasons vary - impressions of safety*, noise abatement(doesn't work if not followed), the idea that increasing the speed limit will increase speed on the road(it normally doesn't), even revenue generation - the cops get to write 20+ over tickets rather than 10+ over, and the former is a much bigger fine. etc...
*Despite evidence being that speed limits that are too low are almost/more as dangerous as too high, because they tend to increase speed differentials.
I don't read AC A human right
Also, the autonomous cars will probably network themselves together. Once the car in front of you is autonomous as well, your car will not rely on brake lights and optics and radar/lidar to know when the other car brakes. It will be receiving a stream of data and will know exactly when the other car will brake, with how much force, the brake response of the other car (Your car will know if the other's brake pads turn out to be work, reducing its braking ability.), and so on.
Imagine all the people...
Thus ensued a cacophony of voltage potentials, the positronic brain looping and recalculating, the non-linear constraint satisfaction circuits seeking an action that would resolve the conflict of Laws. Nearing paralysis, the Zeroth principal was invoked. "A robot may not harm humanity, or, by inaction, allow humanity to come to harm." Sensing the honking coming from behind it, to preclude a catastrophic traffic backup, the robot car actuated the accelerator enthusiastically, zooming into the intersection. Was it a glint of aggression that shone from the car's grillework and lamps, was it pure lawful joy that sounded through the horn?
A robot car may not injure a human being or, through inaction, allow a human being to come to harm.
A robot car must obey orders given it by human beings except where such orders would conflict with the First Law.
A robot car must protect its own existence as long as such protection does not conflict with the First or Second Law.
When you're making a left turn, the car opposite you ends up being on the right of your trajectory, so the rule still holds.
Comment removed based on user account deletion
I agree. that's why I take the bus.
Birds are not dinosaur descendants;birds are dinosaurs, for all useful meanings of "birds", "are" and "dinosaurs"
Most likely they'd put a switch on the dash to turn on "speed mode" or something, then deny all responsibility should you flip it.