Google Self-Driving Car Might Have Caused First Crash In Autonomous Mode (roboticstrends.com)

An anonymous reader writes: While driving in autonomous mode, a Google self-driving car was involved in an accident with a public bus in California on Valentine's Day, according to an accident report filed with the California DMV.The accident report, signed by Chris Urmson, says the Google self-driving car was trying to get around some sandbags on a street when its left front struck the bus' right side. The car was going 2 mph, while the bus was going 15 mph.Google said its car's safety driver thought the bus would yield. No injuries were reported. If it's determined the Google self-driving car was at fault, it would be the first time one of its cars caused an accident while in autonomous mode.

Comments

Might?

[Anonymous Coward]

Not might. It did.

Re:Might?

[Zaelath]

No it didn't.

At 2MPH and 15MPH it's at best a shared fault, and more likely the bus to blame. i.e. you don't have the right to plow into cars you can avoid because they venture into your lane.

Re:Might?

[HornWumpus]

An unsafe lane change would make it the Google cars fault.

The fault is on the vehicle that was changing lanes. Unless they were both changing lanes, it's not a shared fault.

Re:Might?

[Anonymous Coward]

Cars don't get to venture into your lane until it's safe to do so. The bus was in the right here.

Re:Might?

[Zaelath]

Unsafe is a term designed to have flexibility to be determined by a judge. In the same way that "it's never your fault if you're rear ended", it's common wisdom that's incorrect.

Re:Might?

[Lisias]

I don't know about there, but around here if you willingly fails to avoid an accident, you can be charge for life endangerment. The guy that made the mistake would be prosecuted by civil law for the losses, but you would face a prosecution by criminal law.

Unless you are a public transport driver and the victim is a biker. By some reason, they're allowed to ram bikers on the street.... =/ (Don't ask, I don't understand it neither)

Re:Might?

[naris]

Actually, the bus was on the left.

Re:Might?

[avandesande]

You don't know this to be true. A parallel is somebody that swerves into oncoming traffic to miss a squirrel and has head on with another car. I don't know the exact circumstances of the incident, but a bus is hard to control and it is possible that something worse could of been risked by swerving.

Re:Might?

[Anonymous Coward]

Where do you draw the line? What if it is safe unless the car in the other lane floors the pedal and rams into me? What if it is safe as long as the driver in the other car isn't asleep?

Re:Might?

[slinches]

Except neither vehicle was making a lane change. It was a single lane with enough width (normally) to accommodate two vehicles for the purposes of facilitating right turns. In this case, the lane was unexpectedly narrowed by sand bags, so two vehicles attempted to share the lane briefly when there was insufficient space. Fault in these cases is difficult to determine. Technically, since it is still considered a single lane, the bus should not have the right of way. Although, it's likely that the bus driver could not see the obstruction in front of the google car and didn't expect it to move into his path.

There is, however, an argument that a good human driver would have recognized the difference in danger and avoided the incident by just driving over the sand bags. The google vehicle only knew that "something" was in the way and is likely programmed to avoid all round shaped ground obstacles just in case they are a small child or animal.

Re:Might?

[Anonymous Coward]

It's Google's fault, the lane used was wide enough to accommodate 2 vehicles side by side with the far right not being a travel lane but is commonly used in California as a region of the lane used for merging in with traffic or making a turn. Since the autonomous can had to avoid an object in the far right it had to enter the main travel portion of the lane where the bus was and the bud does not need to yield.

Re:Might?

[Anonymous Coward]

There is, however, an argument that a good human driver would have recognized the difference in danger and avoided the incident by just driving over the sand bags.

The safety driver thought the bus would yield. That suggests they wouldn't, and that the safety driver would have had the same crash if they had been driving. Unless they were being complacent and trusted the car knew what it was doing.

Re:Might?

[Anonymous Coward]

If you can't judge when it's safe stay off the road.

If the vehicle is the oncoming lane has to slow down in anyway, it was not safe to enter that lane period.

Re:Might?

[ColdSam]

You would also think that a good human driver would have seen that car going 2 mph trying to avoid the sandbags and not continued on obliviously at 15 mph.

Re:Might?

[Anubis IV]

According to the report, they were side-by-side in the same, double-wide lane, hence the shared responsibility. Here's a Street View picture of the turn in question. Apparently, the sequence was:
1) Red light.
2) Google car signals for a right turn.
3) Google car gets into right side of the double-wide lane and passes cars that are stopped for the red light.
4) Google car has to stop because there are sandbags blocking the storm drain.
5) Light turns green, cars start moving.
6) Google car waits for cars to pass to create an opening, then slowly moves back towards the center of the lane.
7) Bus decides not to yield to the Google car that's ahead of it in the lane, trying to pass it anyway.
8) Bus gets its nose a bit ahead of the Google car.
9) Google car doesn't turn the wheel back in time and scrapes the side of the bus.

More or less, the Google car technically had the right of way, because it was in the same lane as the bus but ahead of it, but the bus had every reason to think the Google car, which was stopped, would cede the right of way to it just the same as it had to several other cars, and thus had clearly decided to pass the Google car. Which is to say, the bus created the situation that caused the impact when it failed to yield to a car that had the right of way, but that doesn't give the Google car a free pass to cause an accident with a vehicle that's already begun passing it. Both had every reason to believe the other would behave differently, yet an accident still occurred, so it's likely a case of shared responsibility.

Of course, most humans would have the common sense to avoid iffy maneuvers around a bus, and the bus driver may have been expecting that as well.

Re:Might?

[jittles]

Unsafe is a term designed to have flexibility to be determined by a judge. In the same way that "it's never your fault if you're rear ended", it's common wisdom that's incorrect.

Unless you intentionally jump in front of someone and then slam on the brakes, how could it ever be your fault if you're rear ended? Even if you make an illegal stop in traffic. And how do you prove that the driver in front was at fault without a video of the incident or unusually good witnesses?

Re:Might?

[jittles]

I don't know about there, but around here if you willingly fails to avoid an accident, you can be charge for life endangerment. The guy that made the mistake would be prosecuted by civil law for the losses, but you would face a prosecution by criminal law.

Unless you are a public transport driver and the victim is a biker. By some reason, they're allowed to ram bikers on the street.... =/ (Don't ask, I don't understand it neither)

That seems like a law that is begging for trouble. How can you possibly determine that someone willfully failed to avoid the accident? If taking action could result in another accident then it is typically better to let the other person hit you. What if dodging that vehicle caused you to crash?

Re:Might?

[Anonymous Coward]

Nope, sorry. Unless the Google AV could change lanes without causing the Bus driver to have to do ANYTHING to modify his behaviour than the AV is at fault...its a classic example of 1 vehicle 'cutting off' another. The only way the Google AV would not be clearly at fault is if a) the bus was slowing down at the time (an indication it 'might' stop) and b) the bus driver was 'waving the Google AV' in (which would raise the question of whether or not the Google AV can recognize a human 'waving them in').

Push comes to shove if you move in to my lane REQUIRING me to hit my breaks to avoid an accident you have NO right to my lane, obviously I may hit my breaks to avoid the accident but this is literally the definition of 'cutting off' someone. This is exactly the behaviour that causes people to 'flip the bird' or honk their horn etc. If you can't make it in to a lane without causing the other vehicle to hit its breaks you shouldn't be cutting in to that lane unless the other driver gives 100% unambiguous indication he has seen you & is 'waving' you in.

Re:Might?

[Zaelath]

You don't need both, either one will do. Just Google "comparative negligence rear-end" there's plenty written about it on law blogs and the like.

Re:Might?

[ColdSam]

By that standard you will be stuck in traffic constantly and be an even greater hazard on the road. The reason this AV even got into this situation was because it was testing software that allowed it to make reasonable assumptions about other traffic as any sane driver would.

Re:Might?

[Darinbob]

Trucks do the same thing at times, assuming that the other vehicle will give way. If a truck wants to merge into your lane then common sense says to move regardless of who has the right to be there. So how do you teach that to an AI?

Re:Might?

[Darinbob]

That's why I really don't like to drive in San Francisco. Impossible to predict what other drivers, pedestrians, or especially cyclists, will do.

As for that particular intersection, you could wait there stopped for hours until someone lets you move over. This is the Bay Area, if you want polite and reasonable drivers you go to Los Angeles instead.

Re:Might?

[mjwx]

If the google car had right of way, traffic laws in your area are fucked up. A car changing lanes does not have right of way. The damage was to the left wheel and front bumper, this means it was hit on the side.

This highlights just how badly autonomous vehicles handle unexpected situations. The driver was expected to intervene at this point but according to the article, he thought the bus would let him in. A defensive driver should know that isn't the case unless the other driver has communicated it to you. They also know a bus will never let you in.

Realistically, the accident occurred because the driver wasn't doing his job.

Re:Might?

[Obfuscant]

it was testing software that allowed it to make reasonable assumptions about other traffic as any sane driver would.

It is exactly this assumption of action on the part of other drivers that AV are supposed to remove from the excuses for accidents, not create more accidents because a computer can't guess what a human is going to do. You're saying that Google is actually making their AV more accident prone.

I can't see how any 'sane' AV would guess that a bus next to it is going to stop before it runs into the side of the bus. No sane human would make such an assumption about any vehicle, much less a bus that is already going by. And I don't think any sane human would guess that someone who had pulled into the right side of a double-wide lane (and was presumably signalling the right turn) would suddenly swerve left back into the main part of the lane. The reason the lane is so wide is to permit easier right turns without blocking straight through traffic -- to assume that everyone who is turning right is going to come back left removes the purpose for the lane in the first place.

Re:Might?

[Zaelath]

Your own reasoning suggests that if a child rushes onto the road between cars, we should charge you with vehicular manslaughter since pedestrians always have the right of way.

The law is not binary.

Re:Might?

[Rockoon]

6) Google car waits for cars to pass to create an opening, then slowly moves back towards the center of the lane.

If it has an opening, then no accident, right?

7) Bus decides not to yield to the Google car that's ahead of it in the lane, trying to pass it anyway.

I see now, the google car was brake checking a 20,000+ lb vehicle by drifting into its path.

People dont do that because they dont want to get hit by a 20,000+ lb vehicle. Google car did because it doesn't know better.

Re:Might?

[Anubis IV]

A car changing lanes does not have right of way.

Had you read what I wrote, you'd have seen that there was no lane change, hence the weird situation. I'm not attempting to defend anyone, just explain based on my reading.

Re:Might?

[dpidcoe]

I think it's there as a catch-all for intentionally hitting someone who just made a minor traffic violation (like in a typical road-rage incident or something). Imagine Florida-man driving his $500 beater truck when somebody in a new BMW pulls out in front of him. Instead of slowing down to avoid it, Florida-man stomps the gas because he technically has the right of way and wants to teach the other driver an expensive lesson.

Re:Might?

[ColdSam]

You are making a lot of assumptions about this incident that aren't warranted. I imagine if I observed your driving behavior for 10 minutes I could find any number of times where you made assumptions about what other drivers were going to do (either that or you would be a tremendous pain to drive around).

But yes, I do think Google is making their cars more accident prone. If they made them completely safe they would be as useless and annoying as the algorithms you claim to use while driving (but I suspect don't).

Re:Might?

[Anonymous Coward]

Trucks do the same thing at times, assuming that the other vehicle will give way. If a truck wants to merge into your lane then common sense says to move regardless of who has the right to be there. So how do you teach that to an AI?

You can teach this quite easy to the AI. It's just like the right of way out on the open ocean. Ships have right of way rules/regs also. They also have one major tweak of the rules, the "Law of Gross Tonnage". Essentially, if the ship is REALLY Big, you give it the right of way no matter what because really big ships can't turn all that well, don't stop quickly and can cause a LOT of damage to YOU and not them if a collision happens. So teach the AI that if the vehicle is above a certain size, let it have the right of way for the passenger's safety.

Gordon

Re:Might?

[tlambert]

Unless you intentionally jump in front of someone and then slam on the brakes, how could it ever be your fault if you're rear ended?

(1) Failure to maintain brake lights. If your brake lights don't go on, then you are not obviously stopping, and therefore you can easily cause an accident.

(2) When antilock braking systems were first introduced, the stopping distance for cars with them got drastically shorter, while the cars not equipped with them kept the old stopping distance. What was previously legally defined as a safe stopping distance was no longer a safe stopping distance for unequipped cars. It's beholden on the person with the shorter stopping distance to take into account the stopping distance of the following vehicle. So lane changing in front of a semi on the freeway and then slamming on your brakes: still not a good idea.

(3) Slow vehicle merges into fast moving traffic. This is a problem, both in terms of lane change merges left (yes, I know: most California drivers are woefully ignorant of traffic laws, because license requirements are so lax compared to other states), but, even more to the point, correct use of acceleration lanes and onramps to get to freeway speeds, rather than getting to freeway speeds only once you are on the freeway.

So: lots of ways to be the guilty party, even if you're the one getting rear ended because you were a dick.

Re:Might?

[tlambert]

Imagine Florida-man driving his $500 beater truck when somebody in a new BMW pulls out in front of him. Instead of slowing down to avoid it, Florida-man stomps the gas because he technically has the right of way and wants to teach the other driver an expensive lesson.

I think I will side with the guy who very obviously has a railroad tie bolted onto the front of his truck as a bumper... I mean, you'd have to be stupid to pull in front of a vehicle like that, even if you weren't doing it to cut them off...

Re:Might?

[Dcnjoe60]

Your description seems to inject blame, particularly with the bus. Maybe a better description would be

1) Red light.
2) Google car signals for a right turn.
3) Google car gets into right side of the double-wide lane and passes cars that are stopped for the red light.
4) Google car has to stop because there are sandbags blocking the storm drain.
5) Light turns green, cars start moving.
6) Google car attempts to slowly moves back towards the center of the lane.
7) Google car scrapes bus that is in center lane.

That is an objective description. It will be up to the authorities to determine who, if anybody was at fault. As for those who are positing that the bus should have stopped. According to the facts it was going 15mph, which means the unrestrained passengers were also traveling 15mph. An emergency stop or swerve would have surely done more harm to them than the sideswipe did, at least using basic physics would indicate it would.

Re:Might?

[Anonymous Coward]

You would also think that a good human driver would have seen that car going 2 mph trying to avoid the sandbags and not continued on obliviously at 15 mph.

You have obviously never met or interacted with a Bay Area Muni driver. If the driver were not stoned or drunk, they would likely have been *aiming* for the car.

Re:Might?

[Kjella]

Trucks do the same thing at times, assuming that the other vehicle will give way. If a truck wants to merge into your lane then common sense says to move regardless of who has the right to be there. So how do you teach that to an AI?

You think this is the first case where Google had to adjust the programming to cope with how people actually drive? Reality is that there's a ton of unwritten rules that don't formally violate any law or regulation but is all about managing expectations, like how we resolve yield deadlocks, lane changing/merging and positioning, passing various obstacles and so on. And yes, large vehicles like buses and trucks seem to follow their own rules sometimes, but Google's car doesn't care what's right or if it got snubbed somehow. If practice suggests it has to yield to the bus, it'll just do that no matter if there's any formal rule that says it should. In fact, didn't they among other things teach the car to break the speed limit by up to 10% to stay with the flow of traffic? I'd love to know what their legal department felt about that, even though it's undoubtedly the right thing to do.

Re:Might?

[Ichijo]

Essentially, if the ship is REALLY Big, you give it the right of way no matter what because really big ships can't turn all that well, don't stop quickly and can cause a LOT of damage to YOU and not them if a collision happens.

More importantly, really big ships must often limit themselves to a very narrow passageway or risk running aground. On the road, this means avoiding low bridges or gross vehicle weight limits, neither of which are relevant in this collision between the self-driving car and the bus.

Re:Might?

[Dcnjoe60]

By that standard you will be stuck in traffic constantly and be an even greater hazard on the road. The reason this AV even got into this situation was because it was testing software that allowed it to make reasonable assumptions about other traffic as any sane driver would.

I'm pretty sure that in every state the law is , if you are blocked in your lane, you either wait for somebody to let you in or you just wait. Just because you are blocked doesn't mean you have the right of way to force your way into moving traffic. If you do that, and I know everybody does, and there is an accident, you get ticketed for failure to yield right of way or improper lane change or something such as that. In effect, you get a ticket for causing an accident, just like the AV did in this case.

Re:Might?

[Anubis IV]

That "objective description" omits a key detail: that the bus was behind the Google car at the time that the Google car began its maneuver. Given that rights of way are determined by facts such as which car was in front at which time, your omission would implicitly place the blame solely on the Google car, when, in fact, the actual facts suggest that the case is more nuanced.

Which isn't to say that my description is objective either, to be clear, nor am I suggesting that the bus should have stopped. I'm merely pointing out that your description has flaws as well.

Re:Might?

[aevan]

It's called 'reversed out into traffic'.

Re:Might?

[ColdSam]

In this specific case it was a shared lane. By what right did the bus get to blindly keep going when there was a car in the same lane that was ahead of it? It has just as much responsibility to avoid other traffic and should only proceed itself when it is safe to do so. Which it obviously wasn't because the Google car was ahead of it. What is reasonable in this case totally depends on where the vehicles were at each point, but the benefit of the doubt should go to the car proceding cautiously at 2mph rather than one going 15 mph and presumably leaving a large gap for some unknown reason.

But in the general case, in just about every merge you are likely to find in a city or busy freeway it will require the trailing car to be reasonable and slow down.

Re:Might?

[perpenso]

Its more general than that. Its a catchall designed to prevent heading into a hazardous situation because you have the right of way. Its a requirement to yield even when other regulations say the other guy should yield to you.

Re:Might?

[LinuxIsGarbage]

Here's someone that basically parked in the passing lane of a highway to help ducklings cross the road. They were charged with criminal negligence causing death when they were rear ended by a motorcycle that killed the riders

Re:Might?

[v1]

3) Google car gets into right side of the double-wide lane and passes cars that are stopped for the red light.

I bet that's what settles it right there. Although it's courteous to do so, with very wide lanes "we assume are wider so we can use them as a storage lane for turning". That doesn't necessarily make it legal to do so. The google car could get ticketed for passing on the right. It could also get ticketed for improper use of parking lane.

Case in point. I was almost in a collision when I came to a stop at a red light at a 4 lane undevided. I had my right turn signal on and was going to turn right, but had to wait for the light to go green because I had one car in front of me. The light turned green, and we all pulled ahead, and I started to execute my right turn, but had to veer left out of my turn and stop suddenly becaue I saw movement in my passenger side mirror.

I narrowly avoided getting T-boned by a car that had come from several positions behind me, driving in the generously wide lane/shoulder, to make a right turn. He didn't see my signal and was assuming we were all going straight, and had reached my bind spot at just the moment I was starting my turn. It's a miracle he didn't hit me and managed to just squeak around me (at a fair speed) to make his right turn. He probably missed my signal because his attention was on the light having just turned green, negating his need to stop before turning, so he was accelerating to the intersection to turn, right as I turned in front of him.

I would expect him to receive a citation for either "passing on the right" or "improper use of shoulder" had we collided. Clearly this wasn't my fault. This isn't far from what happened with the Google car. Had it simply snugged up to the shoulder to turn, and the bus pulled up to his left then, it would be called "passing at an intersection" and the bus would receive the ticket. But in this case, the Google car was driving in a lane that didn't exist, or "passing on the right", and one way or another is probably going to be judged at-fault. This will naturally lead to Google making a change to their driving code to adjust for this possibility.

This raises an interesting question though - if we assume the car believed this was actually a three lane road at the intersection, with the "rightmost lane" being a turn-only lane, then it veering (however slowly) out of its lane and into the other lane where th bus was should be a violation for improper lane change, failure to maintain control, etc. It shoud also have its code tweaked to understand that this was NOT a 3rd lane to be driven in. If it did NOT believe there was an actual 3rd lane, it should not have been trying to drive in it, it's for parking, pulling over, etc, it's a shoulder, not a lane, and driving there without an emergency (or at least with your 4-ways on) is also a voilation.

What should it have done? I don't think it shoud have been trying to go around cars in the shoulder. That would be the first thing I'd fix in the car's coding. Second... how would it have been different if the car HAD snugged up to the right to make its turn, and the bus had then pulled along side to the left. Then we have a problem of the bus technically "passing" the google car, near an intersection, technically a violation unless the bus believed the googe car was exiting traffic and parking. (and most likely, the traffic following the bus doing the same) In that event, the car should stop and wait for all the people to finish their moving violations before it proceeds. It could just consider itself parked and wait to re-enter traffic, which would have the same effect.

It's an interesting situation though - but we all run into a handful of these every year we drive. It's difficult to code for such a wide variety of unlikely scenarios.

Re:Might?

[just another AC]

Of course, most humans would have the common sense to avoid iffy maneuvers around a bus, and the bus driver may have been expecting that as well.

That's a big assumption, and the collision rate of vehicles driven by irrational bags of meat begs to differ.

I have seen many drivers with "they will yield to me because their car is more expensive and I don't care about mine" or "I can make that" or "f**k them I'm in a rush" or any other number of ego driven bad choices.

Yes a rational thoughtful person might act as you said, but they are a rare specimen nowadays.

Re:Might?

[Lisias]

How can you possibly determine that someone willfully failed to avoid the accident?

By analyzing the accident. There's enough space on the road to avoid the hit? At what speed was you? At what speed was the culprit?

If you were speeding, you was on reckless driving.

If there was space to avoid the hit, you at very minimum was incompetent to void the accident and should not be driving.

Your right to be right is not greater than the right of other to be not be injured if you can avoid the injury.

If taking action could result in another accident then it is typically better to let the other person hit you.

Being that a good and valid excuse to the accident. All you have to do now is to prove your point - what you must do it anyway, what would you do if the other guys accuses you of being responsible by the accident?

What if dodging that vehicle caused you to crash?

What if by preventing your car to crash, a kid is dead?

You are the kind of driver that, if a kid crosses the street without looking, you prefer to hit the kid than a tree? Of course the three will trash your car but you would only be seriously injured if you would be speeding - what makes you negligent at best.

Re:Might?

[perpenso]

A car changing lanes does not have right of way.

Had you read what I wrote, you'd have seen that there was no lane change, hence the weird situation. I'm not attempting to defend anyone, just explain based on my reading.

If there is a "double wide lane" to accommodate right turns then there is essentially an implicit right turn lane despite the lack of paint on the road. The google car was attempting to maneuver around an obstacle by moving from the implicit right turn lane to the implicit traffic lane.

I've studied the CA DMV handbook a few times over the decades and I don't recall any such thing as a "double wide lane". However I do recall something about space reserved for parkings spots and bicycle lanes can turn into right turn lanes at corners.

Re:Might?

[Lisias]

I think I will side with the guy who very obviously has a railroad tie bolted onto the front of his truck as a bumper... I mean, you'd have to be stupid to pull in front of a vehicle like that, even if you weren't doing it to cut them off...

Unless the other guy's vehicle is a bus or truck weighting 10 times your car's weight - in which event your neck will receive full impact, as the railroad tie will not deform and will transfer fully the impact to the passengers. ;-)

Re:Might?

[LinuxIsGarbage]

It was a single lane with enough width (normally) to accommodate two vehicles for the purposes of facilitating right turns. In this case, the lane was unexpectedly narrowed by sand bags, so two vehicles attempted to share the lane briefly when there was insufficient space. Fault in these cases is difficult to determine. Technically, since it is still considered a single lane, the bus should not have the right of way.

Based on the street view it looks like it's primarily that wide to facilitate the ability to park, and drive, in the same lane. By the same note if a parked car pulls out in front of a bus is it still the bus's fault? The report mentions it is an articulated bus. That means it's even heavier and less maneuverable than a normal bus.

The article mentioned the car signaled the intention to turn right, however there was no mention whether it canceled the signal, or signaled left to indicate the desire to move into the center portion of the lane. This could be confusing / conflicting information for the bus driver. A stopped (or virtually stopped) car in the right side of the lane, near an intersection, with a right turn signal, is not expected to move left.

Re:Might?

[Ichijo]

I narrowly avoided getting T-boned by a car that had come from several positions behind me, driving in the generously wide lane/shoulder, to make a right turn. He didn't see my signal and was assuming we were all going straight, and had reached my bind spot at just the moment I was starting my turn.

That means either the other car drove off the road, or you weren't far enough to the right. In California, what you did would be a violation of CVC 22100(a):

Both the approach for a right-hand turn and a right-hand turn shall be made as close as practicable to the right-hand curb or edge of the roadway

Also in Arizona:

Both the approach for a right turn and a right turn shall be made as close as practicable to the right-hand curb or edge of the roadway.

The two are almost word for word identical. Your state probably has the same law. If I were the judge, you would both be charged, you for violating the above law, and the other driver for reckless driving or unsafe pass.

Re:Might?

[Anonymous Coward]

You have obviously never met or interacted with a "INSERT AREA" Muni driver. If the driver were not stoned or drunk, they would likely have been *aiming* for the car.

There, fixed that for you. They will ALL run you over given a chance.

Re:Might?

[dgatwood]

That's pretty screwed up. In the U.S., she might get a ticket, but the cyclist would be found to be 100% at fault for the wreck. Speed limit on the roads in the U.S. are, by law, required to be set such that a vehicle moving at the limit would have adequate time to stop even if there's a vehicle stopped or other obstruction on the road, and that's probably true in Canada as well. If a vehicle approaching from behind fails to stop, regardless of the reason why the vehicle fails to stop, that vehicle should always be at fault unless the front vehicle shifted into the lane in front of another car and then immediately slammed on the brakes.

I find it particularly mind-boggling that the judge found that the motorcyclist's grossly excessive speed was not a significant factor. The motorcyclist was going 80 in a 55 zone. That's 145% of the posted speed limit. The motorcyclist would have gotten automatic jail time for that sort of gross recklessness in most of the U.S., had he survived. More to the point, had he been traveling the speed limit, he would have had almost twice as long to slow down, and likely would have been able to dodge the car entirely.

As for the Google situation, legally, it likely depends on how far back the bus driver was when the Google car started around. With that said, San Francisco bus drivers are (or so I'm told by friends who are crazy enough to actually drive there) notorious for not stopping for cars stopped in bus lanes. Google's car needs to make the assumption that Muni buses never yield the right of way even if they legally should. Anything less is just inviting an accident. :-)

Re: Might?

[Anonymous Coward]

As a fellow biker, the biker that got himself and his daughter killed was not doing his job, and it sounds like he was likely speeding as well. If you're scanning ahead for 12 seconds, things like this are easy to avoid. Lane changers and arbitrary left makers on the other hand... Not so much.

Re:Might?

[Anonymous Coward]

"This is a lighthouse... your call."

Re:Might?

[Obfuscant]

By what right did the bus get to blindly keep going

Who said it was "blindly"? The car ran into the bus, which means the car had a much better view of the impending accident than the bus driver did. And even if the bus driver did see it coming, was he supposed to slam on the brakes or swerve into the next lane over to avoid it?

when there was a car in the same lane that was ahead of it?

You mean the car in the right side of a wide lane, with a right turn signal flashing, that was stopped until the bus was next to it? That car "in the same lane"?

only proceed itself when it is safe to do so. Which it obviously wasn't because the Google car was ahead of it.

Yes, I'd agree, when there is a Google car ahead of you, it is not safe to proceed.

but the benefit of the doubt should go to the car proceding cautiously at 2mph rather than one going 15 mph

Why? The car that pulled over into a bus is not at fault for pulling over into a bus? The car that was going so slow that it would have taken almost zero time to stop when it realized the bus was next to it, compared to a bus going 15 MPH, likely with passengers that had no seat belts? The bus didn't rear-end the Google car, the Google car pulled over into the bus. I think the benefit of the doubt goes to the vehicle doing the normal, regular operation, not the one doing something unusual.

Re:Might?

[Zaelath]

That's exactly my point, jackass.

Re:Might?

[justthinkit]

What you missed is that some roads -- typically Interstates -- require vehicles to maintain a certain speed (hence the "no farm vehicles" signs). I would say that 0 mph was an insufficient speed.

Re:Might?

[ColdSam]

Once again you are making a lot of unwarranted assumptions to fit your own biases against Google and AV cars in general. You are bringing nothing of value here and it is not worth my time (or that of anyone else serious about this) trying to have a reasonable conversation with you.

Re:Might?

[Zaelath]

No darling, my point was all these "well understood laws" are not as well understood as most people think, and certainly not as clear cut.

In the same way as a series of jackasses are convinced the Google car is automatically 100% at fault, the "pedestrians have the right of way" fallacy would mean anyone striking one dead is guilty of manslaughter 100% of the time.

Maybe read for context and get that burr out of your ass, or you know, just stop being a cunt.

Re:Might?

[Anonymous Coward]

If you're 50m behind me coming in at 15mph, and i change into your lane, you will be forced to stop. The law only prevents dangerous driving, it doesn't guarantee that you won't have to stop. You can't just go an and ram me because it's your lane, if you do so you're the one breaking the law. Shut up and apply your brakes.

Re:Might?

[Anonymous Coward]

So, there's a un-marked right turn lane, Google Car (let's call him 'Gar') is in it. Then Gar decides something is in the way so he wants to move back into the normal straight lane. In the meantime, Mr. Bus is driving down the straight lane - ignoring those idiots turning right because they aren't in his way - when Gar runs into him.

The main fault here is that there is an unmarked right turn lane. Either 1) it shouldn't be used for right turns AT ALL by anyone, or 2) it should be marked.
With it not being marked, things like this happen.

Re:Might?

[AthanasiusKircher]

What you wrote implies that the car had pulled into a de facto lane or parking area or perhaps a shoulder of the road. The car was far enough to the side to be out of the main traffic lane and for other cars to pass -- that's what you said. So regardless of whether it was a marked lane or not, the car was effectively pulled off in another de facto lane. Cars reentering traffic from a stop after pulling off to the side also don't have the right of way, regardless of whether they're in another lane, on the shoulder, whatever.

Re:Might?

[Rockoon]

No darling, my point was all these "well understood laws" are not as well understood as most people think, and certainly not as clear cut.

When you declared something to be a law that wasn't even close to a law, your point was that "well understood laws" are not as well understood as most people think?

Let me fix this for you:: "well understood laws" are not well understood by you.

You are not most people and therefore you do not get to decide what it is that most people understand, and certainly not purely based on your own lack of understanding.

Re:Might?

[v1]

He was off the road, driving on the shoulder. It was a four lane undivided road, and I was stopped in the rightmost lane, 2nd from the front. He originated from many cars back, and chose to leave the right lane and drive along the right shoulder, to bypass around a dozen cars that were in that lane waiting for the light. He probably figured that since the line wasn't moving, the person at the front of the lane (at the intersection) wasn't turning right or would have done so by now, so he was free and clear to jet around the stationary cars and get turned sooner.

But as he was approaching the intersection, the light turned green and the one car that was ahead of me (preventing me from turning right on red) departed. So I pulled ahead and tried to begin my turn, just as this other fellow got there. He was voilating several traffic laws, I was violating none and had no reason to expect or even suspect anyone coming up from behind me on my right. It was only my peripheral vision and good reaction time that allowed me to get out of his way before he ran into me as I turned in front of him.

Re:Might?

[ColdSam]

A parked car is a very different situation. A reasonable driver should be aware of those turning right in the shared lane and should proceed with caution. A parked car without a turn signal jumping into the lane is much harder to account for.

Re:Might?

[Anonymous Coward]

From the article, the car plowed into the bus, not vice versa.

Re:Might?

[fluffernutter]

In my mind, if you are driving down a road and there is an obstacle, you STOP IMMEDIATELY if it is not save to pull around it. I must side with the bus on this one. The only direction a driver has the 'right' to drive is straight, and only if there is no one going slower directly in front of them.

Re:Might?

[Dcnjoe60]

That "objective description" omits a key detail: that the bus was behind the Google car at the time that the Google car began its maneuver. Given that rights of way are determined by facts such as which car was in front at which time, your omission would implicitly place the blame solely on the Google car, when, in fact, the actual facts suggest that the case is more nuanced.

Which isn't to say that my description is objective either, to be clear, nor am I suggesting that the bus should have stopped. I'm merely pointing out that your description has flaws as well.

That is true, the bus was behind the google car while the car was blocked from advancing. Then the light turned green and the bus and the rest of traffic started moving and the the google car tried to switch lanes and turned into the side of the bus, sideswiping it. The blame is solely on the google care because the bus did nothing wrong.

Re:Might?

[ColdSam]

Absolutely right. Fortunately the other vehicle wasn't a kid on a bicycle. This is a prime example of why we can't get autonomous vehicles soon enough.

Re:Might?

[JoelKatz]

Did you read the accident report? The car wasn't changing lanes. It was moving from the right side of its lane into the the center of its lane. The bus, moving much faster and overtaking the car from behind, apparently attempted to share the lane with the car.

At least, that's what the accident report, filled out by Google, claims.

Re:Might?

[Ichijo]

He was off the road, driving on the shoulder.

Was it a paved shoulder or unpaved? The answer to this is important in determining whether the shoulder was part of the roadway.

Re:Might?

[Zaelath]

Your own reasoning suggests that if a child rushes onto the road between cars, we should charge you with vehicular manslaughter since pedestrians always have the right of way.

Ok, one last bite. When did I suggest that /any/ of this was a factual statement? You just cherry-picked a sub-sentence out of a post, out of a broader context, and then ridiculed the result.

In fact, it was an example of how people misread and misunderstand the law, and in fact as you say make stupid shit up all the time, pass it around, and think that repetition makes it fact. Worse, some of those people are police officers.

You go right ahead and continue your rant, it's entertaining. Doesn't matter if your shouting at the right target or not.

Re:Might?

[fluffernutter]

You feel it would have been better if the bus swerved away from the car, thus entering the oncoming traffic lane?

Re:Might?

[ColdSam]

Is a head on catastrophic collision killing dozens the only other possibility you can envision?

We don't have a full video of the accident, so we can't judge with any certainty. However, there are dozens of things that the bus driver could have done differently. To start with it shouldn't have even been in the situation where it had to act suddenly because it shouldn't be accelerating past a car so close to it that was clearly blocked and might move over a few feet. It also shouldn't have left such a large gap from the car in front of it if it didn't want others to merge. Lastly even if letting off the gas and slowly braking wasn't an option, yes going a foot into the next lane most likely would have been fine.

Re:Might?

[Ken D]

So... just plow into the stopped vehicles in front of you? In order to maintain speed?
Obviously you've never been in stop and go traffic on an interstate.

Re:Might?

[Anonymous Coward]

If the Google car never turned its right blinker off, I wouldn't have expected it to move left too.

Re:Might?

[ColdSam]

Your point was understood all along by anyone with an ounce of sense and no agenda. You are wasting your time here.

Re:Might?

[Anonymous Coward]

There is, however, an argument that a good human driver would have recognized the difference in danger and avoided the incident by just driving over the sand bags.

Maybe, maybe not. But I'm sure we can all agree that your *average* driver would have gunned it while pulling in front of the bus, all while laying on the horn and shouting obscenities.

Re:Might?

[Anonymous Coward]

I'm confused. If he was driving on the shoulder (which is wrong) and he nearly hit you, then you must have been driving on the shoulder (which is wrong) or just about to do so.

Re:Might?

[William Baric]

I obviously don't live in the same part of the world as you.

1) Although I agree it's the responsibility of the driver to make sure his brake lights work properly, you should never rely on brake lights of others to determine if someone is braking or not.

2) It's not the responsibility of the driver in front of you to brake slowly because your car sucks. It's you who have the responsibility to keep a safe distance considering your car.

3) Even if the car merging come to a dead stop in front of you (it frequently happens where I live because of sudden traffic jams), it's your responsibility to be in control of your car. If you rear end him, then chances are you were not.

Basically, where I live we think the driver is responsible for his car. I know, we're weird.

Re:Might?

[Hognoxious]

The key word being "roadway". Does that include the shoulder or not? GP doesn't seem to be sure either.

Re:Might?

[Zaelath]

Of course I'm wasting my time; This is Slashdot! /kicks Persian into a Poll

Re:Might?

[ArylAkamov]

This. It sounds like the car let several others pass, then when the buss was half way past it decided to move and turn into the buss.

I don't see how the buss would be at fault.

Re:Might?

[dgatwood]

What you missed is that some roads -- typically Interstates -- require vehicles to maintain a certain speed (hence the "no farm vehicles" signs). I would say that 0 mph was an insufficient speed.

Although that is true, that offense is typically an infraction (at least in the U.S.). By contrast, exceeding the speed limit by 25 MPH (80 in a 55 zone) would normally be a misdemeanor. If both parties committed some sort of traffic violation leading to a wreck, you'd expect blame to fall upon the driver whose violation was more serious.

To add further justification for that opinion, in this case, there's no practical difference between stopping to help ducklings and stopping because of a sudden traffic backup, an animal in the road, or a wreck. If traffic been stopped for any of those reasons, the motorcyclist would still have died because his grossly excessive speed did not leave him adequate time to stop when he found traffic suddenly stopped around the corner. So IMO, the car driver should have been issued a citation for an infraction for illegally obstructing the highway, but she should not have had any culpability in the death of the motorcyclists. That culpability should fall squarely on either the motorcycle driver (if the road's speed limit was reasonable) or the government (if a review found it to be too fast based on the limited visibility around that corner).

Re:Might?

[mysidia]

how could it ever be your fault if you're rear ended?

When you are stopped, and a vehicle is a few feet behind you, throw your car into reverse, and accelerate: bonus points if your reverse lights are non-functional.

By the way, when the police comes to take a statement, say you were stopped at the light and got hit from behind --- be sure not to recall having put your car in reverse.

Re:Might?

[aevan]

From the article, that was the upper end estimated by the police. 70~80 [113~129 kmph] was the estimate, and at least in the Ontario highways (yes, I know that it's Quebec, but they are even worse) it is extremely common for traffic to flow at 15km over the speed limit. Of course, the wife of the motorcyclists claims they were only doing 85 kmph.

She also didn't even bother to put on the hazard lights when she stopped in the passing land, simply stopped and hopped out - not even closing her door. Additionally, other drivers nearly struck the parked car, saying they had no ability to stop but could only swerve. She definitely deserves negligence for deciding to park in the middle of a highway.

Re: Might?

[aevan]

If the timeline for the incident is as tight as the the various papers had the eyewitness accounts, there was another vehicle between him and the parked, no-lights car. The driver of the vehicle in-between barely swerved out of the way in time upon noting the parked car, then stated she saw another vehical (the motorcyclist) didn't clear it. That distraction of her swerve could have drawn the motor cyclists attention just long enough to prevent him from swerving himself.

As for speed, police estimate speeding of of 20 to 30kmph over the limit (no idea what speed the vehicle that swerved was doing, but unlikely to been slower than the motorcyclist else the cyclist would have been up in her muffler - so assuming 'keeping with flow'), while the wife is saying 'we were doing 5kmph under the speed limit because I'm a nervous driver but we were traveling together as a family'.

Re:Might?

[ColdSam]

Nobody is sure, actually. Every driver and every cop will give you a different answer as to what is safe and expected in such situations. The rule for non-jerks and honest officers is just for everyone to be reasonable. In this case it means the poster should probably have been further to the right, clearly indicated he was turning right and looked for any traffic in the shoulder. Of course, the other guy also made many errors and shares fault in this near accident.

Re: Might?

[Anonymous Coward]

Learn to check your mirrors, you self righteous prick. Even when something "shouldn't" be there.

Re:Might?

[Jack Griffin]

No it didn't.

At 2MPH and 15MPH it's at best a shared fault, and more likely the bus to blame.

Of course it is... keep telling yourself that...

Re:Might?

[N1AK]

I see now, the google car was brake checking a 20,000+ lb vehicle by drifting into its path.

Keeping with /. tradition and jumping to entirely unsubstantiated hyperbole I see. You don't see shit, you've imagined up a scenario based on very little information that could be anything from accurate to completely different from reality.

Re:Might?

[Jack Griffin]

Unsafe is a term designed to have flexibility to be determined by a judge.

Yet here you are, blaming the bus...

Re:Might?

[Jack Griffin]

Of course, most humans would have the common sense to avoid iffy maneuvers around a bus, and the bus driver may have been expecting that as well.

Absolutely, which is why there is more to driving a car than simply knowing programming a rulebook into the computer.
In Thailand, if a foreigner is in a car accident, it is their fault regardless of circumstances. The reasoning is if you weren't in the country, the accident wouldn't have happened. A similar reasoning could be applied here, if a human was driving, regardless of rules they wouldn't try it on with a bus, and no accident would've happened. Hence the robot is at fault.

Re:Might?

[Jack Griffin]

So teach the AI that if the vehicle is above a certain size, let it have the right of way for the passenger's safety.

You haven't really thought this through. The end result of this logic is that any hill-billy in a pick-up will simple charge on through anytime they see a nerd in a Google-mobile. While that may be a technically safer option it isn't sustainable. The unwritten rules of the road is that sometimes you have be prepared to hold your ground. I can't imagine that any AI capable of that will be legal.

Re:Might?

[ColdSam]

That is an insane analogy. Making an arbitrary decision on who you choose to blame doesn't change who actually is at fault.

Re:Might?

[jandersen]

As for the Google situation, legally, it likely depends on how far back the bus driver was when the Google car started around

Speaking of which, an autonomous car would have have an array of sensors monitoring things around the car, and one would expect that the data will have been logged for exactly this sort of eventuality, so it ought to be easy enough to see who was in violation of the rules.

Re: Might?

[Anonymous Coward]

Piles of case law disagrees with you. ALL jurisdictions in the US and UK (I've researched it and IAAL) disagree with you outright on point 1 from a road regs point of view.. You're also wrong according to case law on 2 and 3. I'm less verse with the US case law on these matters and more so on mainland Europe.

Unless you live in some bizarre jurisdiction in West Africa or something, you're wrong.

Re: Might?

[Anonymous Coward]

I think you have just identified some missing features from the Google car. Automatic insults and random rapid acceleration accompanied with aggressive maneuvers. Then Google could claim honestly that their cars were at least as safe as the average human driver

Re: Might?

[Anonymous Coward]

Even if that became a chronic problem, I cannot imagine that the national Highway system will become grid locked because a couple of hicks drive crazy.

Re:Might?

[Dog-Cow]

You are either easily confused, or you've never made a right turn from a road with a shoulder.

Re:Might?

[driblio]

Again, why did other drivers have 'no ability to stop'? That's entirely their fault.

Re: Might?

[Anonymous Coward]

ABS did not reduce stopping distances. It made them longer but permitted you to safely steer around obstacles.

Your comment highlights the complete lack of understanding in an average driver. No wonder there is demand for autonomous vehicles.

Re:Might?

[stealth_finger]

There is, however, an argument that a good human driver would have recognized the difference in danger and avoided the incident by just driving over the sand bags.

The safety driver thought the bus would yield. That suggests they wouldn't, and that the safety driver would have had the same crash if they had been driving. Unless they were being complacent and trusted the car knew what it was doing.

Google guy obviously hasn't been driving long if he thinks a bus is going to give way, ever.

Re:Might?

[Zaelath]

Unsafe is a term designed to have flexibility to be determined by a judge.

Yet here you are, blaming the bus...

And in a different post...

You haven't really thought this through. The end result of this logic is that any hill-billy in a pick-up will simple charge on through anytime they see a nerd in a Google-mobile. While that may be a technically safer option it isn't sustainable. The unwritten rules of the road is that sometimes you have be prepared to hold your ground. I can't imagine that any AI capable of that will be legal.

...here you are suggesting that the problem with the AI is that it can't legally be impatient like a human, implying you would have went in front of the bus same as the safety driver.

Re:Might?

[Anonymous Coward]

Maintaining speed is not for safety - it is to avoid clogging up a good road with slow tractors. Cars sometimes stop abruptly due to faults. There may be debris that fell off some truck. There may be stopped cars due to an accident or even a hole in the ground. You are not allowed the assumption that the road is clear ahead.

  In car-motorcycle accidents it is often the car driver's fault - but this was clearly the rider's fault. Lots of people might not care so much about the ducklings - but what if she stopped for some person walking in the middle of the lane?

Re: Might?

[Sarten-X]

In an emergency situation, ABS allows the brakes to be kept more firmly on while maintaining control, compared to manual pumping of the brakes while maintaining control. The ABS brakes have a higher duty cycle than manual braking, so they can effectively stop the vehicle in a shorter distance.

Re:Might?

[Rob Lister]

That is true, the bus was behind the google car while the car was blocked from advancing. Then the light turned green and the bus and the rest of traffic started moving and the the google car tried to switch lanes...

That's the nuance previously referred to. Technically it did not try to switch lanes. It tried to move to the left side of the lane it already occupied. According to California law ...

Pass traffic on the left. You may pass on the right only when: An open highway is clearly marked for two or more lanes of travel in your direction.

The driver ahead of you is turning left and you do not drive off the roadway to pass.

Never pass on the left if the driver is signaling a left turn.

https://www.dmv.ca.gov/portal/...

By that time the google car was signally left.

The bus should have yielded.
OTOH, never, ever cut off a bus. Which the google car did.

Re:Might?

[tehcyder]

So lane changing in front of a semi on the freeway and then slamming on your brakes: still not a good idea.

Yes, but that's out of self preservation, not because of the law.

Re:Might?

[KozmoStevnNaut]

Google guy obviously hasn't been driving long if he thinks a bus is going to give way, ever.

Yeah, the informal Rule of Gross Tonnage applies every bit as much on the road as on the sea.

Re: Might?

[RavenLrD20k]

Responding as a Georgia Driver very well versed in Georgia Driving laws and have successfully self represented in court on several occasions. Thus, IANAL... but I don't need one on this.

First, a link to someone who is a Lawyer that represents in Georgia, USA

Point 1. The Driver of Car A (the lead car) involved in a rear-end collision will receive a Citation for faulty equippment, however will not be cited at fault for the accident. That will fall on the driver of Car B (the trailing vehicle).

Point 2. The case law is clear on this one. You must be able to perform an emergency stop and not collide with the vehicle in front in the event the vehicle in front of you performs an emergency stop of his own for ANY given reason. Whether this be he's avoiding an emergency stop in front of him, or avoiding a collision with a creature, it doesn't matter. It is the trailing driver's responsibility to maintain a safe buffer, whatever is safe for the conditions at the time, no ifs ands or buts.

Point 3. This is the only point your parent is wrong about. A driver that changes lanes and suddenly decelerates, thus creating an unsafe condition in which a collision occurs and he is rear-ended, that driver will be considered at fault for the accident and receive a citation on scene for "Improper Lane Change".

Again, I am not a lawyer, but I will play one legally and successfully in traffic court more often than not. My only argument with you is that I took exception to your blanket statement of "ALL jurisdictions" and gave you one concrete example of the exception that makes your statement patently false.

Re:Might?

[RavenLrD20k]

This wasn't stop and go traffic. This was an open interstate where suddenly there was a car stopped in the "hammer lane," while traffic was moving at normal speed around her. The lane is so named because it's the left hand lane where speeds are the highest.

Re:Might?

[cb88]

From a practical standpoint the larger vehicle always wins... pulling out in front of a moving bus in tiny car is nothing short of stupid.

Re:Might?

[jittles]

How can you possibly determine that someone willfully failed to avoid the accident?

By analyzing the accident.., you was on reckless driving.

Oh hey! Look at that! There is already a law that covers driving like an idiot.

If there was space to avoid the hit, you at very minimum was incompetent to void the accident and should not be driving.

Your right to be right is not greater than the right of other to be not be injured if you can avoid the injury.

If there is room to stop, what person would fail to stop? But I am not going to jump into another lane to avoid hitting someone just because they cut me off. That could result in a far more serious accident. And I have the right to avoid doing things that could cause harm to someone else. If someone does something that puts me and them in harm, then that is their fault, not mine. At least, in a general sense. There are always cases where this may not be 100% true. Taking precipitous action would just put someone else at risk and not the person who made the first mistake.

If taking action could result in another accident then it is typically better to let the other person hit you.

Being that a good and valid excuse to the accident. All you have to do now is to prove your point - what you must do it anyway, what would you do if the other guys accuses you of being responsible by the accident?

Why should I have to prove that I did not commit a criminal act? It seems to be that the government should have to prove that I did. Whether or not I am at fault is a civil matter, not a criminal matter.

What if dodging that vehicle caused you to crash?

What if by preventing your car to crash, a kid is dead?

You are the kind of driver that, if a kid crosses the street without looking, you prefer to hit the kid than a tree? Of course the three will trash your car but you would only be seriously injured if you would be speeding - what makes you negligent at best.

Oh come on now. This is not a car on person accident. This is a car on bus. While more likely to be lethal than a car on car accident, all occupants were in a steel cage designed to protect them in the event of a crash. A car on motorcycle/bicycle/pedestrian situation is NOT comparable to a car on bus accident. No one in their right mind would intentionally hit another human being to avoid damaging their paint job on a tree. If someone did intentionally hit a human being then, again, there is already a crime for that! It's called (attempted) murder if they did it on purpose and manslaughter if it was an avoidable accident.

Re:Might?

[Anonymous Coward]

This wasn't a lane change. It was shifting from the right of the lane to the middle of the lane. One lane. Vastly different.

Re:Might?

[Dcnjoe60]

A vehicle in a moving lane of traffic is not required to yield to a vehicle trying to enter that lane, unless the vehicle entering the lane is an emergency vehicle, of course. The bus did exactly what it was supposed to do. Could it have stopped or swerved to avoid being hit, possibly, but if you take that approach than all of those collisions where a google vehicle was rear ended because it reacted quicker than the drivers behind it should be google's fault because it failed to adjust to the traffic conditions. It would seem to go both ways.

The reality is that the bus did nothing wrong and was required to do nothing different. At 15mph a sudden stop or swerve could have seriously injured passengers (who are not restrained). Chances are that if the bus was an AV and the car wasn't, because of the potential harm to the passengers, it would have done the exact same thing.

However, this story is old now and not really worth discussing any longer. AV have great potential to improve many areas of life, but like everything else, they cannot guarantee 100% safety and never will. To be successful, they only need to be "more" safe than regular cars (and at a price point that most people can afford).

Re:Might?

[bluefoxlucid]

In practice, right-of-way alternates in this situation. If the lane is clear and you proceed, you have right-of-way. If the traffic coming the opposite way plows through, they are violating your right-of-way. Whether it's actually safe to proceed in such a fashion is another matter; it's *reasonable*, it's *legal*, and it's really the only way to handle the situation besides just parking and waiting for the police to send an officer to direct traffic, but it's not necessarily safe.

Re:Might?

[bluefoxlucid]

Wait, it wasn't approaching from the opposite direction? In my state, if you signal to change lanes while moving with traffic, the driver behind you must yield; failure to do so is a $250 fine. Assholes around here like to speed up into your blind spot to keep you from changing lanes in front of them, hence the fine.

Re:Might?

[bluefoxlucid]

Technically, sideswiping another vehicle at 2mph isn't unsafe; it's just inconvenient.

Re:Might?

[Anubis IV]

Wait, it wasn't approaching from the opposite direction?

Nope, they were both in the same lane heading in the same direction (at the time), though the Google car was signalling that it would take a right turn, so the bus was trying to pass the Google car (while staying in the lane) on the left. When the Google car swung a bit to the left to re-center itself in the lane, it sideswiped the bus that was passing it.

Re:Might?

[virtual_mps]

An unsafe lane change would make it the Google cars fault.

And a safe lane change (signaling, plenty of room) followed by a distracted driver not noticing that there was a car in front would be the bus driver's fault. Now that we've established the basics, we'll have to see what actually happened instead of speculating wildly.

Re:Might?

[fluffernutter]

Well that's the point of AI isn't it? What would a human consider reasonable and how would they resolve the situation safely. Then do that.

Re:Might?

[Talderas]

There is a practical reason. Nothing about the situation required that she bring her vehicle to a stop in a lane of traffic let alone the passing lane. She could have brought her vehicle to a stop on the side of the road. She let common sense be overridden by her animal loving sentiment. Her actions, negligent as they were, created the hazardous situation that resulted in the death of the motorcyclist. In the US she could have easily been charged with involuntary manslaughter.

The other examples you cite, except the animal one which the dead animals that are hazards are rare, are all hazardous situations that would not have been of her creation or due to negligent behavior on her part.

Re:Might?

[rhazz]

I've had similar circumstances though never such a close call yet, mostly because I am familiar with the intersection and know to watch for it. On my daily commute there is an intersection with a right-hand turning lane which becomes available about 100 feet from the intersection. The roadway itself has a wide paved shoulder which is clearly marked. Often traffic backs up from the red light, and people further back who want to turn right will drive down the shoulder to access the turning lane - I've seen people do at least 100 feet on the shoulder, passing other cars like my own already signalling the turn but waiting for proper clearance. In general I don't think it is a bad thing when traffic is very slow and the people doing this go slowly and carefully, because it does get them on their way sooner and out line up going straight. But this is an 80 kph road and the occasional asshole doesn't slow down while doing it.

Re:Might?

[JargonScott]

Seems like it needs a function of "if opposing vehicle is X times larger than itself" then just capitulate.

Re:Might?

[bluefoxlucid]

Ahhhh, that's a problem. It should have signaled to the *left* to indicate its lane change, even for a partial lane change.

Re:Might?

[Anubis IV]

I don't disagree that it implies exactly what you said. Whether or not that is technically and legally correct, however, I don't know. Someone else was mentioning that there are laws on the books in California regarding these sorts of lanes where there's a shoulder/parking/turn lane and how they are to be used. If there are indeed, then I imagine that it will play out as you said. If there aren't any applicable laws, then things get a bit hairy.

Re:Might?

[Triklyn]

there was a car in front that swerved out of the way.

and with shitty visiblity, and blockers like bends and hills... i've been surprised by a love seat in the middle lane of 3 lane before. going 65mph, that shit will make your heart go. just barely swerved, and i imagine if there was a car behind me even at a nominally safe distance, there would have been a collision.

when i got to my destination 10 minutes later, i called that shit in.

because stationary objects in the middle of a highway, are a fucking accident waiting to happen.

Re:Might?

[Triklyn]

no blinkers on. no brake lights either. parked was parked.

Re:Might?

[Anonymous Coward]

That was my first thought having grown up in California myself, buses don't give a crap about other vehicles or cyclists. All of my scariest encounters with side-swipes were buses trying to cut me off to make a right turn in front of me when I was cycling and had the right of way through the intersection.

Re:Might?

[hucker75]

Bus drivers don't obey the rules of the road, take this clown for example: https://youtu.be/LJzCLq55nvg

Re:Might?

[ChrisMaple]

There's more than just a speed consideration for prohibiting farm vehicles on public roads. Some farm vehicles have steel wheels that tear up roads.

Re:Might?

[Obfuscant]

Once again you are making a lot of unwarranted assumptions

And your claim that the bus driver blindly ran over the Google AV is even more of an unwarranted assumption.

You seem to ignore the fact that the bus driver probably had a bus load of passengers to worry about, and making an emergency stop would probably put them either on the floor or potentially through the front window. He didn't have time to think about whether he had straphangers or not. He did know that the car had people in seat belts and other safety equipment, so any injuries to them would be extremely minor, if there were any at all (and there weren't.) The correct response to a car that is pulling over in to him is to not slam on the brakes, not swerve (like the AV was doing to get around the partial obstruction), but to brake in a controlled manner to a safe stop, and let the repairmen fix the damage.

You're also repeatedly ignoring the fact that the bus did not swerve over to hit the car, the car swerved over to hit the bus. The car hit the side of the bus, not the other way around.

As for assumptions, I think it is a very safe assumption that the AV was signalling a right turn, at least until a fraction of a second before it pulled left, and it would be impossible for anyone to determine that the right turn signal was no longer on for at least a second. (Even when it is on, it spends a large amount of time unlit, you know.)

What I don't believe is an assumption is the fact that to the bus driver, the AV looked like a car that had pulled over to the curb and stopped -- because that is what it had done.

You are bringing nothing of value here

All you bring is the AV Uptopianism that ignores the problems that this accident demonstrates. The car "assumed" someone else was going to do the right thing in the split second there was to make a decision and it ignored the fact that a very large vehicle was already in the place it was trying to go.

You're ignoring the fact that the car was programmed to ignore the difference in speeds and stopping times, to ignore that it could stop a lot faster than the bus, that it could stop a lot more safely than the bus, and it chose to take the right of way that the laws (of man) say it might have had. It ignored the laws of physics in preference to an artificial law.

The AV, in this case, chose to act like an arrogant human driver, pushing its way back into moving traffic, into a vehicle that it could not win a game of chicken with, assuming that the other guy would flinch and give way.

trying to have a reasonable conversation with you.

If your repeated "you are making assumptions" is the only argument you have, then there has been no reasonable conversation with you.

Re:Might?

[ChrisMaple]

The google car was going 2 mph. It could have stopped in inches. It should have stopped, and waited for traffic to clear. Going over sandbags might not be physically possible, and attempting to go over them would likely have damaged them, opening the driver for charges of deliberately damaging government property.

Re:Might?

[JimFive]

If the line on the edge of the road is solid you are not supposed to cross it, anything past the solid white line is not part of the roadway.

Re:Might?

[ChrisMaple]

The bus was going 7 times as fast as the google car. The bus sideswiped the google car, not vice versa.

Re: Might?

[Anonymous Coward]

So, in other words, from point 3, the Google self-driving car was at fault (if this took place in Georgia).

Re:Might?

[Anubis IV]

Speed has nothing to do with who sideswiped whom. It's a matter of who maneuvered into the other. In this case, the bus was going straight while the Google car moved into the side of the bus. Thus, the car sideswiped the bus, albeit, at slow speed. Otherwise, by your logic, if I'm being passed by a car and enter their lane as they nose ahead of me, it's them who sideswiped me since they were going faster than me, even though I'm the one who maneuvered into them.

Re:Might?

[michaelwigle]

I proved it by pointing at the 10 yard fresh skid mark that started at the lane marker and moved toward the middle of my lane. Driver in front told the officer they thought they saw something in the road just as they changed lanes (right in front of me). Officer cited them for something like erratic driving (I don't really remember). It does happen, although probably not too often. Which is why the conventional wisdom says if you do the rear ending you're at fault.

Re: Might?

[RavenLrD20k]

Yes, that would be correct. It is the driver's responsibility to ensure that there is space enough for his vehicle to make a lane change without causing the potential of a collision. The "blind spot" defense is not valid.

As part of my insurance company's driver education materials, if it comes down to the choice of a comprehensive claim (Deer strike, significant pot-hole, hit a tree, wind up in a ditch) or a collision claim (hit another vehicle or a guard rail), aim for the comprehensive if possible. Then you'll be less likely to face legal penalties as well as it'll be cheaper for the insurance company since there's only damage/medical to pay out for the insured's vehicle as opposed to however many vehicles are involved in the collision.

In this case, it seems the Google car made the wrong choice either due to lack of information (sensor was not aligned to cover potential blind spots or calibrated improperly) or calculation error. Also note that the summary states that the Google car was not rear-ended, but instead hit the bus in its side with the front of the Google Car. This indicates that the bus should have entered the car's "danger detection zone" before the car began moving into the lane. Again, the car didn't calculate something correctly, whether it be that it didn't receive all the necessary information or a bug in the program flow.

Disclaimer: This is only based on information as reported. The official investigation may turn up mitigating factors that I haven't accounted for that may alter my perceived judgement.

Re:Might?

[Anonymous Coward]

Then the person wandering in the lane would have been charged with manslaughter or murder, among other things, for the biker's death.

Re: Might?

[Anonymous Coward]

"compared to manual pumping of the brakes while maintaining control."

But that's not what happens.

Normal drivers mash the brakes, the wheels lock and they lose control.

Re: Might?

[Anonymous Coward]

"The ABS brakes have a higher duty cycle than manual braking, so they can effectively stop the vehicle in a shorter distance."

No.

A non-ABS braking system on "full" with locked wheels is the "best case" for stopping distance.

Comparing a non-ABS braking system with a human driver "pumping" the brakes in some kind of way, as opposed to ABS pumping the brakes in a very definite way, is an impossible comparison to make.

The ABS programming should represent the maximum level of braking while retaining control of the vehicle. A good human driver should be able to recreate this, or err on the side of caution and maintain control while sacrificing stopping distance. The reality is that most drivers are not good and will simply lock the brakes.

Re:Might?

[Anonymous Coward]

So.

You are driving a 2ton (or more) lump of metal at high speed. You do not have a right of way, nor should you ever assume that the road ahead is clear. If you CAN'T SEE your stopping distance then you must slow down until you can. Guess what that means? Slowing down for corners.

Too many dickheads drive to the friction limit of their tires, as though this is somehow safe.

Everyone should drive to the limit of their stopping distance, or the tire friction, or the posted speed limit, whichever is lowest.

And if their is an obstruction up ahead, you simply stop the fucking car. There is no need for such a senseless accident.

Re:Might?

[Anonymous Coward]

You're fucking crazy.

There can *never* be the assumption that the road ahead is clear. To do so is madness.

No wonder the US road toll is so damn high.

Re: Might?

[Anonymous Coward]

How about, the car in front reversed into me! I was stationary at the time.

Re:Might?

[Anonymous Coward]

Actually the nautical right of way rules are governed by maneuverability.

Once you start to look into it you must take into account things like navigation channels (a large ship may be unable to leave the channel due to draft) and cable barges (entirely un-maneuverable but appears to be so) or a fishing vessel towing nets.

This is why a motor boat gives way to sail, the motor board is more maneuverable than the sailing vessel, which is limited in its ability to turn and move relative to the wind.

I sailed for two years, and I think it's made me a more considerate and a better driver. You have to do 10 (100?) times more thinking when in control of a vessel at sea. Can only imagine what 1,000,000 things the aviators have to think about while coming in to land.

Re: Might?

[tlambert]

LOL

OK

"I have one question, father... how fast was the bastard going when he backed into you?"

Re:Might?

[ColdSam]

This is why you (and everyone else) need to be replaced by an AV.

You make absolute statements about how crashes should not be avoided if there is even a chance of causing a different accident, whereas those you are arguing with are making the point that you should make the best decision at the time: sometimes that means taking a low risk on another accident just to avoid a high risk or certainty of an accident.

When given examples you quickly change your original answer that, of course, you wouldn't run over a kid, that would be crazy.

You are incapable of making such decisions even while sitting at your computer, so it is unlikely you will make good decisions in the split second before an accident. You would have slammed into the Google AV instead of going a foot over into oncoming traffic because that might have caused another accident. Ignoring of course that the certainty of slamming into the AV might cause it to careen onto the curb killing that child that you now seem to be concerned about.

Re:Might?

[Xolotl]

Technically there was only one lane which the Google car was already in.

Re:Might?

[dgatwood]

That negligence was factored into the law when they decided to make blocking traffic an infraction, not an offense that can result in jail time. If you want to treat the negligence as somehow more important than that, then you would have to make a claim that a reasonable person would expect blocking traffic to have that sort of outcome in this particular circumstance.

Vehicular manslaughter generally requires recklessness, which usually requires that the person be aware of the likelihood of stopping causing a fatal collision and exhibiting gross disregard for that risk. At least in my mind, no reasonable person would expect that stopping a car on a highway would cause a motorcyclist to hit the back of his or her car at 70 MPH, particularly if there's a second driving lane. Most people are aware of their surroundings, and travel at a speed that isn't grossly above the speed limit, and as such, most people assume that other drivers will do the same. A driver has to make not one, but several serious errors to end up in a 70 MPH fatal collision with a stopped vehicle.

Re:Might?

[delt0r]

Of course, most humans would have the common sense to...

Where the fuck are these amazing humans everyone here keeps talking about? Because they are not here in the real world. Just work at an insurance company on claims for one week.

Humans are crap drivers and do not generally or even specifically exhibit any such "common sense".

Re:Might?

[Lisias]

Why should I have to prove that I did not commit a criminal act? It seems to be that the government should have to prove that I did. Whether or not I am at fault is a civil matter, not a criminal matter.

Because, besides being innocent until proved guilty, no one has the duty to prove your innocence - and you can bet your sorry ass that the guilty part would profit by accusing you.

If the other guys accuses you of reckless driving, you better prove you weren't.

Re:Might?

[Jack Griffin]

There is nothing arbitrary about it. On the road there exists the actual road rules, and animal instinct. It doesn't matter if the road rules say I have right of way, If I see a bus barrelling towards me I'll do whatever is on my power to avoid it. AI can't have that programmed into it's logic and still remain legal.

Re:Might?

[Jack Griffin]

...here you are suggesting that the problem with the AI is that it can't legally be impatient like a human, implying you would have went in front of the bus same as the safety driver.

Er no, better go back and read it again...

Re:Might?

[KozmoStevnNaut]

Well yes, the rule of gross tonnage is a simplified way of looking at it, obviously there are some caveats, as you mentioned.

I'd like to echo your experience that sailing has made you a more considerate driver. After I started riding motorcycles, I found that I became a much more courteous and patient driver. It's all about breaking of of your comfort zone and trying something other than the normal car you've become complacent in.

Re:Might?

[v1]

The road was paved, and the shoulder unpaved. It was a gradual enough and wide enough shoulder to comfortably drive on though, which encourages people to pull around stopped traffic to make a rifght turn, so it happens frequently at that intersection and other similar intersections around town.

You most often see it when there's only a few (or one) car stopped in the right lane and someone right behind them wants to turn right, and they only go around one or two cars and turn. This was unusual because there was such a long line of cars that it made it impossible for me to see him in my rear view mirror and it limited his view of my turn signal. Then there was his assumption that no one was turning right because the line wasn't moving. His driving down the shoulder at a fairly high speed only made matters worse.

The light turning green and allowing me to pull forward to make my turn broke his assumptions. And yes, I do keep peripheral view of my mirrors. But no, I don't specifically stsop to check my right mirror for a car driving off-road to try to go around me, that's not a reasonable expectation for any driver. Drivers need to keep the majority of their attention ahead of them, where cars are expected to be. The only thing that belonged where it was would have been an emergency vehicle, which is authorized to use the shoulders. (with caution and care) And in that case, the flashing lights would easily have gotten my attention much earlier.

A poster a few down seems to think it's my responsibility to watch out for idiots breaking the traffic laws, "get out of their way", to prevent collisions. No, that's the idiot's job. People need to accept responsibility for their actions, and quit expecting me to take responsibility for their (or anyone else's) actions. I'll pay attention to what I'm doing, and you pay attention to what you're doing, and we'll all be fine. It's like these parents that think the whole world needs to watch out for their kid because they failed to teach their kid common sense. But thats another argument for another time.

Re:Might?

[mjwx]

A car changing lanes does not have right of way.

Had you read what I wrote, you'd have seen that there was no lane change, hence the weird situation. I'm not attempting to defend anyone, just explain based on my reading.

Had you read the article, you would have seen that there was a lane change.

The damage happened on the front side quarter, that means the car was hit at an angle as it crossed the path of another vehicle.

Re:Might?

[Anubis IV]

The damage happened on the front side quarter, that means the car was hit at an angle as it crossed the path of another vehicle.

Correct.

Had you read the article, you would have seen that there was a lane change.

Not so. And I did read the article.

Again, as I've already explained, there was no lane change, since the Google car never exited the lane it was in. It did exit the flow of traffic while remaining within the lane, which is why it was then able to cross the path of another vehicle, but, once again, it did so without ever technically changing lanes (i.e. no lane markings were ever crossed), hence why this is a weird situation.

Re:Might?

[jittles]

This is why you (and everyone else) need to be replaced by an AV. You make absolute statements about how crashes should not be avoided if there is even a chance of causing a different accident, whereas those you are arguing with are making the point that you should make the best decision at the time: sometimes that means taking a low risk on another accident just to avoid a high risk or certainty of an accident. When given examples you quickly change your original answer that, of course, you wouldn't run over a kid, that would be crazy. You are incapable of making such decisions even while sitting at your computer, so it is unlikely you will make good decisions in the split second before an accident. You would have slammed into the Google AV instead of going a foot over into oncoming traffic because that might have caused another accident. Ignoring of course that the certainty of slamming into the AV might cause it to careen onto the curb killing that child that you now seem to be concerned about.

No I am saying that there is no way to legislate that someone 'avoid an accident.' Not that you should not avoid an accident. I'm saying that you should not make erratic lane changes to avoid another car. This is what the entire thread has been about. The google car made an erratic move, at a speed significantly slower than the flow of traffic, to avoid hitting a sandbag. If you hit a sandbag what is it going to do? Bend the control arm on that wheel? Mess up your fender? Perhaps. But you didn't exacerbate the situation. If someone cuts you off and your options are to make an unsafe lane change or hit the person in front of you, you should do your best to slow down and stay in your lane. We were never discussing plowing into a child until you brought it up. Avoiding killing a child is a natural reaction that no one would fault you for. I would fault you for causing an accident with another vehicle when you're trying to avoid damage to your own vehicle. My point has always been that you cannot legislate that someone avoid an accident. Once you do that, people are going to think they're required to swerve or otherwise do stupid things to avoid fault in an accident. There are already laws that cover negligent driving. You want people making safe decisions and not erratic ones because they're worried about criminal penalties for failing to avoid an accident. That has been my argument the entire time. I am sorry you failed to grasp that.

Re:Might?

[jittles]

Why should I have to prove that I did not commit a criminal act? It seems to be that the government should have to prove that I did. Whether or not I am at fault is a civil matter, not a criminal matter.

Because, besides being innocent until proved guilty, no one has the duty to prove your innocence - and you can bet your sorry ass that the guilty part would profit by accusing you.

If the other guys accuses you of reckless driving, you better prove you weren't.

Well hold on here. Reckless driving is a criminal matter. He can claim reckless driving all he wants but without some sort of moving violation or other criminal charge against me how can he possibly bring that accusation up in court? There has to be some evidence to prove this - whether it is eyewitness or physical. Your word against theirs isn't going to count much either way without other evidence. I realize the standard of evidence is much lower in a civil court but my entire argument has been against criminal penalties for failing to avoid an accident. The civil case for damages is a whole different ballgame.

Re:Might?

[Anonymous Coward]

A reasonable person would actually expect the possibility to be rear-ended when fully stopped on a highway *when the other lanes are moving at high speed*. This actually does happen frequently enough, be it due to lack of visibility, inattentive drivers, curved roads, someone swerving at last moment to uncover the blockage in front of you, etc.

It is extremely risky to even remain in your car or too close to it, because of that.

Re:Might?

[Anonymous Coward]

Well, at least LA drivers can be expected to be polite while reloading. :)

Re:Might?

[ColdSam]

What you are failing to factor in to your argument is math. Mostly logic and probability, but a little geometry too.

It is insane to say that you should always run into the car/person/animal, ... rather than take evasive action, which is what you are saying. That may by the safest choic, but it is not always the safest choice. Further, of course, the government can and should make laws saying that people should make reasonable efforts to avoid accidents. The law should fault the person who runs into the car or child in front of them, rather than swerving slightly onto the shoulder on an otherwise completely open road. As discussed above, it may be difficult to convict someone of this, but the law should encourage them to do the safest thing which is NOT just plowing into the car in front of you ignoring everything else around.

Re:Might?

[Lisias]

Well hold on here. Reckless driving is a criminal matter. He can claim reckless driving all he wants but without some sort of moving violation or other criminal charge against me how can he possibly bring that accusation up in court?

Here.

And also here

Re:Might?

[ColdSam]

This is nonsense. Neither the law nor any reasonable person would say that such a bus always has the right of way. But this absolutely does NOT mean that a reasonable person wouldn't still yield that right for self-preservation and an AV can certainly be programmed with the same logic. Just as it could yield at an intersection to ANY car that appeared to be not slowing down or running through.

Re:Might?

[Lisias]

My point has always been that you cannot legislate that someone avoid an accident. Once you do that, people are going to think they're required to swerve or otherwise do stupid things to avoid fault in an accident. There are already laws that cover negligent driving.

And our point is that you are plain wrong. You don't even know how the laws you mention works.

A lot of your arguments is plain wishful thinking.

You want people making safe decisions and not erratic ones because they're worried about criminal penalties for failing to avoid an accident

Wrong. You want people avoiding accidents, no matter how. It's the whole purpose of the law: to AVOID ACCIDENTS.

That has been my argument the entire time. I am sorry you failed to grasp that.

And I'm sorry for you being you. It must be difficult living this way, pulling out wishful thinking out of the ass and then pretending be sorry by the other people "failing" to follow your nonsense. :-)

Re:Might?

[ColdSam]

Let me get this straight. Do you think that stopping your car in the middle of a multi-lane freeway where traffic is flowing at full speed is a safe thing to do? Because that sure seems to be what your statements imply.

I don't know of any reasonable person who would do such a thing except in an extreme emergency. Could they guess the exact type and speed of vehicle that will run into them? Probably not, but they will realize that they are probably in the most dangerous situation they have been for a very long time.

Re:Might?

[Hognoxious]

Thinking about this, I've never seen a shoulder other than on freeways, where you don't usually get intersections.

No doubt KGIII knows of some. Probably his idea to put them there.

Re:Might?

[ColdSam]

You have changed this story so as to make it almost unrecognizable to the original that you were comparing to the Google AV accident. The car you are now describing was clearly reckless and breaking many laws and there is just no comparison anymore. It's now just a story about his terrible driving (and your bad driving).

As it stands it still sounds as if you were not as far to the right as you reasonable could, you were probably in the middle of the lane aligned with all the other traffic and therefore making it hard for anyone else to realize that you were going right. (If you were as far to the right of the roadway as possible then you should have mentioned that long ago.)

The main reason for this law (and good safety practice) is because it is quite legal for bicyclists (even stupid kids) to ride to the right of that line of cars EVEN on the shoulder. The law, which you seem to either disrespect or ignore, is to help prevent such traffic from getting crushed when you suddenly veer right from the middle of the lane without checking your side mirrors.

Re:Might?

[dgatwood]

I didn't say it was a good idea. It disrupts traffic and greatly increases the odds of minor accidents because of vehicles merging to pass you, not to mention posing a significant risk of a minor, low-speed crash if drivers don't notice the vehicle in time to fully stop.

However, a reasonable person wouldn't expect a stopped car to cause a 70 MPH collision in a 55 MPH speed zone. You'd expect drivers to be starting out at the speed limit +/- 5 MPH rather than +25 MPH. You'd expect them to slow down, merge, and avoid the obstacle like a competent driver would, rather than slam right into the obstacle like somebody who isn't paying the slightest bit of attention to the road. And so on. That sort of accident requires gross negligence on the part of the other driver.

More importantly, had the motorcycle been traveling at the speed limit (55 MPH) instead of 80 MPH, the cycle would have been going at most 40 MPH at the time of impact, and probably slower. At 40 MPH, the odds of a double fatality are about one in 200. At the 70 MPH estimated impact speed, the odds of a double fatality rise to about one in four. That's a factor of fifty difference, and that's without factoring in the fact that the motorcyclist likely could have avoided the crash entirely at a slower speed by merging into the other lane, so it is probably more like a factor of several hundred difference.

Based on that, the car's driver is less than 1% responsible for those deaths.... That's stretching the definition of negligent manslaughter pretty badly, IMO.

Re:Might?

[ColdSam]

Why are you trying so hard to avoid the question? We all agree that it is not a good idea, that was not the question. Is it safe?

You seem to continue to think so and you are quite out of touch with reality on this one. Why do you keep stressing the particular details of the accident as if you think that matters? All that matters is that this was a very dangerous (i.e. not SAFE) and irresponsible act that could reasonably be foreseen to cause an accident. That combined with an actual fatality is all that matters. We could argue over whether the actual fatality should matter, but our legal system seems to think so.

Do you also think it's safe to fire a gun at random in a room full of people? I probably only have a 1% chance of killing someone, so that makes it "safe" and reasonable to you? If it happens to hit and kill two people then that is even more unlikely so I am even less responsible, by your reckoning.

What matters in both these cases is that you are taking an action that could reasonably be expected to cause harm or death. It doesn't matter that death is not a certainty and you don't calculate the odds and say it only had a 1/10 chance of killing him so I am only 1/10th responsible. This is craziness.

Re:Might?

[Jack Griffin]

This is nonsense. Neither the law nor any reasonable person would say that such a bus always has the right of way.

Except this law here you mean?

But this absolutely does NOT mean that a reasonable person wouldn't still yield that right for self-preservation and an AV can certainly be programmed with the same logic.

Yet here we are, google car crashing into a bus....

Re:Might?

[ColdSam]

You have no idea what you are talking about. The link you provide doesn't support anything you're saying and you clearly didn't even read the whole thing (if you are even able to) or you would have seen why. Did you just google "bus right of way law" and post the first dumb thing you found? Looks like it.

Peddle your idiotic ideas elsewhere, there is absolutely nothing to be gained from talking to you anymore (if there ever was).

Also, start training for new employment, because the only explanation for your irrational fears of AVs is you are worried they will put you out of a job.

Re:Might?

[Jack Griffin]

Wow. You could've chosen to debate actual points but instead you went full retard...

Re:Might?

[dgatwood]

Why are you trying so hard to avoid the question? We all agree that it is not a good idea, that was not the question. Is it safe?

Is it safe? Of course not. It is just orders of magnitude safer than racing down the highway at 80 in a 55 zone.

Do you also think it's safe to fire a gun at random in a room full of people? I probably only have a 1% chance of killing someone, so that makes it "safe" and reasonable to you?

That falls clearly under reckless disregard for human safety. Drawing a gun in a crowded room, even if you don't shoot it, puts people's lives in danger. Even before you shoot, people near you are likely to trample each other trying to get away. And if you do shoot, such a reaction is almost assured. So there's almost a 100% chance of serious injury or loss of life if you draw a gun in a crowded room and shoot it, even if you deliberately shoot it in a way where the bullet can't hit anyone.

Stopping on a road is orders of magnitude less dangerous than pulling out a loaded gun in a crowded room. Cars stop on a road all the time, and the overwhelming majority of the time of the time, the biggest harm it causes is causing a traffic backup that makes people late to work. To put the risk in perspective, the closest thing that we have statistics for are secondary accidents, where one traffic accident causes a second accident in the backup. The probability of an accident in a traffic backup is somewhere on the order of 0.5%–1.5%, and that's all secondary accidents, including minor fender-benders where somebody cuts into somebody's side while changing lanes at 2 MPH in the backup.

Now, here's where we multiply. About half of one percent of accidents result in a fatality. When we multiply that out, stopping your car has about 25 ten-thousandths of one percent chance of killing someone. So statistically speaking, you'd expect, on average, a little over two deaths if you stopped your car in a driving lane 100,000 times, give or take. That's why deaths in secondary accidents are so rare that you can't even find statistics for them.

And that's based on actual accidents, where you have multiple vehicles stopped in a lane for half an hour or longer, as compared with a single vehicle stopped for a couple of minutes. So that probably reduces the odds of a fatal crash by another order of magnitude. In other words, this really was approximately a one-in-a-million long shot.

So no, no reasonable person could possibly expect that stopping on the road would cause a death, much less multiple deaths. You're five times as likely to die from a lightning strike over the course of your life than to kill someone by making a single illegal stop in the middle of a busy freeway. From a safety perspective, that makes such a stop only slightly unsafe, not grossly negligent (which is the standard that you would need to prove if you wanted any sort of vehicular manslaughter charges to stick in the U.S., typically, to the best of my understanding).

Re:Might?

[v1]

You should read the law before attempting to apply it.

CVC 21755: "The driver of a vehicle may overtake and pass another vehicle upon the right only under conditions permitting that movement in safety."

("vehicle" applies to any transportation operating on the road, and 21755 is most often applied to bicycles, and bicyclists must be keenly aware of this responsibility for their own safety, but the rule is the same for cars as well)

In other words, the onus of maintaining safety is entirely on the passer. That'd be the guy that almost hit me because he wasn't paying attention. And in the google case, the google car became a passer on the right while attempting to drive around cars to get into position for the right turn. Then merging back into traffic again required the google car to take on the burden of safely merging, which it failed to do. The google car was either passing on the right or exiting and merging back into traffic, depending on how you want to look at it, and either way it was entirely its responsibility to insure safe navigation with existing traffic. (If you're going to try to reply with an argument/exception against the above written law, be sure you cite your source, I'd like to read it)

Re:Might?

[Agripa]

Unless you intentionally jump in front of someone and then slam on the brakes, how could it ever be your fault if you're rear ended?

Changing a tire while stopped in the center lane of a freeway at night would count. This happened to a friend of mine.

Re:Might?

[ColdSam]

Excellent point. If anyone who is not an idiot cares to get back to me I will address this.

Re:Might?

[ColdSam]

Excellent point. If anyone who is not an idiot cares to get back to me I will address it.

Re:Might?

[JimFive]

Rural state (and federal) highways have shoulders and intersections.
--
JimFive

Re:Might?

[Anonymous Coward]

I come from an alternate universe where the lady ignored the ducks. The motorcyclist is still alive, but he is in a coma and quadriplegic. He hit one of the ducklings at high speed and lost control of his vehicle due to his excessive speed.

The approach is too hard...

[Anonymous Coward]

Trying to build a car that can drive itself and cover all cases is too hard. You need to start small and slowly add more and more safety features to a vehicle before it can handle most of the difficult scenarios of accidents.

Re:The approach is too hard...

[mikael]

It's like being at driving school again. Learn to start the engine, ensure the engine is in neutral, put the car into 1st gear, accelerate, brake, learn to turn left and right, reverse driving, reverse turning, going up and down hills (lower and higher gears). Then you try various obstacles, more complicated road layouts like hidden junctions, roundabouts, exits hidden behind hedges and trees, traffic lights on steep hills, hairpin junctions that require sharp turns. Various hazards like school zones, one way roads, farm vehicles and animals, riding schools, wild animals in forests.

Re:The approach is too hard...

[bondsbw]

It's hard to conclude that from a fender-bender in a situation in which humans make the same type of error all the time... and also considering that both parties shared some responsibility.

Talk to me when the number of injuries or fatalities approaches even 1/10 that of human drivers. Until then, I want to see how this plays out.

Re:The approach is too hard...

[Anonymous Coward]

Talk to me when the Google car drives even 1/10 as much as all those millions of human drivers, driving ten times as much each day. And when Google finally decides to even start testing this greatest invention of the 21st century in places other than California, in weather other than perfect. And when it doesn't drive like an 80 year old, holding up everyone else on the road.

Re: The approach is too hard...

[Anonymous Coward]

For the number of miles driven it is significantly safer when compared against an average person driving the same. I would also suspect that we would see even greater safety in areas other than California, have you ever driven here?

Re:The approach is too hard...

[Anonymous Coward]

You need to start small and slowly add more and more safety features to a vehicle before it can handle most of the difficult scenarios of accidents.

Which is exactly what they've been doing for the past several years.

Re:The approach is too hard...

[wisnoskij]

You do realise that /. had an article a month ago about how driver-less cars have been involved in a statistically higher number of accidents so far? In fact the statistic stated was twice as many accidents than the average human driver.

Re:The approach is too hard...

[Anonymous Coward]

Hopefully no one remembers that. In 50 years we're going to have a bunch of people sitting around whining about where their driverless ca is that they were promised.

Re:The approach is too hard...

[bondsbw]

Yet only one accident, the one in the article, is the fault of the automated system. That seems to suggest that the cars are not yet as capable to help prevent humans from causing accidents, but then wouldn't that suggest that a road full of them would be safer?

Regardless, this tech is still in the pre-alpha phase. Dismissing something that is already showing very promising results, something that isn't even close to being on the market, doesn't seem productive.

Re:The approach is too hard...

[tehcyder]

Regardless, this tech is still in the pre-alpha phase.

Then, with the greatest respect, it shouldn't be on public roads.

Re:The approach is too hard...

[bondsbw]

Learners permits allow pre-alpha drivers on the road.

30 of the 24 bus passengers to sue Google

[NotDrWho]

Ow my neck!!

Re:30 of the 24 bus passengers to sue Google

[Fly Swatter]

30 of the 24 bus passengers to sue Google

Do you happen to tally votes for a living ?

Re:30 of the 24 bus passengers to sue Google

[Anonymous Coward]

This. In Philly, the driver would refuse to open the doors to prevent a rush of people ON to the bus.

Re:30 of the 24 bus passengers to sue Google

[__aaclcg7560]

Do you happen to tally votes for a living ?

OP must work for Wall Street. The stock market had predicted nine out of the last five recessions.

Re:30 of the 24 bus passengers to sue Google

[Anonymous Coward]

30 of the 24 bus passengers to sue Google

Do you happen to tally votes for a living ?

Works for Hillary

Re:30 of the 24 bus passengers to sue Google

[Anonymous Coward]

Works for the hilarity known as trump?

Re:30 of the 24 bus passengers to sue Google

[Anonymous Coward]

Pleez. Philly bus drivers take pride in the number of people who can be uncomfortably stuffed into a bus.

Re:30 of the 24 bus passengers to sue Google

[Attila Dimedici]

There was a case of an out of service (not carrying any passengers) Philly bus which was in an accident, by the time the police arrived every seat was occupied.

Re:30 of the 24 bus passengers to sue Google

[Anonymous Coward]

nek minnit

Re:30 of the 24 bus passengers to sue Google

[Rockoon]

Works for Hillary

Only in 7 of the 57 States that Obama visited in 2008.

Re:30 of the 24 bus passengers to sue Google

[Anonymous Coward]

At least Obama has visited states, some of the candidates haven't even heard of Baja California.

Re:30 of the 24 bus passengers to sue Google

[penguinoid]

Just wait a bit and you'll see -- the bus driver will admit it was his fault. Also, he'll be reporting a large "gift" income on his tax return this year.

Re: 30 of the 24 bus passengers to sue Google

[Anonymous Coward]

Baja California translates to Lower California, or in Drumpfese, Loser California. They're going to pay for that wall, all right? LOSERS!

Machine Learning

[sunderland56]

>> Google said its car's safety driver thought the bus would yield.

So Google is teaching their cars to drive like normal Californians: expect that the other guy will yield.

Re:Machine Learning

[Anonymous Coward]

Please read the statement again. The safety driver thought the bus would yield, they did not say anything about what the Google AI expected the bus to do.

Re: Machine Learning

[Anonymous Coward]

The Google AI was already driving in the wrong lane. So this is clearly 100% safety driver fault

Re:Machine Learning

[nukenerd]

The safety driver thought the bus would yield, they did not say anything about what the Google AI expected the bus to do.

Looks like that is what the AI thought too.

Re:Machine Learning

[Anonymous Coward]

Please read the statement again. The safety driver thought the bus would yield, they did not say anything about what the Google AI expected the bus to do.

Well, either the autonomous system expected the bus to yield or it didn't see the bus.

Re:Machine Learning

[BasilBrush]

It was in AV mode. Therefore the fact that it was moving means the AI thought the bus would yield also.

Re:Machine Learning

[Anonymous Coward]

The safety driver was expecting the bus driver for one to salute his robotic overlords.

Re:Machine Learning

[Theaetetus]

>> Google said its car's safety driver thought the bus would yield.

So Google is teaching their cars to drive like normal Californians: expect that the other guy will yield.

It's a good point... Living in Boston, I've learned to never expect buses or taxis to yield. It's like they drive with their eyes closed half the time.

Re:Machine Learning

[ColdSam]

They're certainly not doing themselves any favors in justifying not having their jobs taken away by AVs.

Re:Machine Learning

[Anonymous Coward]

expect that the other guy will yield.

That's not what happened.

Here's the short story of the actual events.

Google car was in the far right-hand lane and approaching an intersection, and was going to turn right.
A short ways before the intersection, that lane 'flares' out... i.e. it gets wider. But it does not split into two lanes.
The Google car used a recently added algorithm, which had it pull closer to the curb while preparing to turn, as opposed to hugging the inside stripe (closer to the center of the road). The idea is that it gives cars behind you the ability to pass as you turn right, so that traffic doesn't stack up.
The Google car detected sandbags around a storm drain, blocking its path, so it decided to move back to the left side of the lane to get around.
At this point it detected the bus, as did the Safety driver, but since there was still plenty of room they both assumed that the bus would not illegally pass in the same lane. But apparently neither of them understands that bus drivers are quite frequently psychotic assholes, and the bus driver instead accelerated forward on the left side of the lane, so the Google car bumped into the bus.

No word yet on if the bus driver was cited for failing to yield to a vehicle ahead of it or for illegally passing within the same lane.
Google has adjusted the algorithm to assume that large vehicles are less likely to yield when they're supposed to.

Buses have right of way

[presidenteloco]

in some jurisdictions, cars have to yield right of way to buses in general.
Buses certainly have right of weight.

Also, what's with the aggressive / obnoxious sneaking around cars in same lane tactic. Did someone program that or did the software learn it?

Re:Buses have right of way

[hey!]

That may be true in some jurisdictions, but what's true in all jurisdictions is driving is that right of way isn't a license to get into an accident that you can avoid. If the Google car really was traveling at just 2 mph, then you have to wonder whether the bus driver could have avoided the accident.

In any case it's clear that if the safety driver had been driving the accident still would have happened; he judged that the bus would yield, but it didn't.

Re:Buses have right of way

[Anonymous Coward]

That's proprietary info, you are not going to have access to it until after a couple of gigantic court cases involving dead children.

CAPTCHA: unfair

Re:Buses have right of way

[newcastlejon]

At 2mph I wouldn't be surprised if the bus driver thought the Google car was stationary.

Re:Buses have right of way

[jittles]

That may be true in some jurisdictions, but what's true in all jurisdictions is driving is that right of way isn't a license to get into an accident that you can avoid. If the Google car really was traveling at just 2 mph, then you have to wonder whether the bus driver could have avoided the accident.

In any case it's clear that if the safety driver had been driving the accident still would have happened; he judged that the bus would yield, but it didn't.

I do not believe that most jurisdictions require you to take action to avoid someone else hitting you. That could result in far more dangerous circumstances. And if the google car was going 2mph then the correct action is for the google car to stop for the sandbag rather than jump in front of a bus. Besides, I thought the whole point of autonomous cars is that they're supposed to be safer? It is clear that the car merged into the bus.

Re:Buses have right of way

[Obfuscant]

And if the google car was going 2mph then the correct action is for the google car to stop for the sandbag rather than jump in front of a bus.

From the description in the summary, it sounds like the car ran into the side of the bus. It didn't jump in front of it, it sideswiped it as it tried to go around sandbags in its lane. Assuming that the bus was in its own lane, the car had to leave the lane it was in to do that.

Every discussion about safe driving I've seen in this forum has had the "safe" drivers claiming that the only safe thing to do is stop when faced with an impediment to traffic, not to try swerving around it. And the autonomous discussions have all claimed that AV will be safer than humans. Considering that the car was going just 2 MPH, I cannot imagine that it could not have stopped for the sandbags and waited until the next lane was clear, but I'm sure someone can explain why it didn't.

And the car "safety driver" assumed the bus would yield to him. I for one welcome our new robotic overlords.

Re:Buses have right of way

[Tablizer]

At 2mph I wouldn't be surprised if the bus driver thought the Google car was stationary.

Good point. Sometimes overly-cautious drivers are confusing because they act different than "normal" drivers.

We are tuned to expect a certain degree of human jerkativity and aggressiveness.

But that's not a reason to ban auto-cars, it just means the "culture shift" will have bumps in the road (pun intended), as is typical with technology change.

Re:Buses have right of way

[Darinbob]

The good part is that the Google car has a whole lot of telemetry data waiting to be analyzed to figure out what really happened.

Re:Buses have right of way

[Anonymous Coward]

TFS is misleading.

Read TFA.

Re:Buses have right of way

[LynnwoodRooster]

And if the google car was going 2mph then the correct action is for the google car to stop for the sandbag rather than jump in front of a bus.

From the description in the summary, it sounds like the car ran into the side of the bus. It didn't jump in front of it, it sideswiped it as it tried to go around sandbags in its lane. Assuming that the bus was in its own lane, the car had to leave the lane it was in to do that.

See, that's the thing - they were in the same lane. The AV was in the right side of the lane preparing to turn, the bus was behind and starting to pass the AV on the left side of the lane. The AV saw sandbags in its way, so slowed and moved over - and the bus did not yield. You cannot pass another vehicle in its same lane in nearly any situation (the one I know of that is legal is lane splitting on a motorcycle in California). The bus was in the wrong - it was passing another vehicle in the same lane.

Re:Buses have right of way

[Obfuscant]

the bus was behind and starting to pass the AV on the left side

To hit the side of a bus, it has to already be passing you.

The bus was in the wrong - it was passing another vehicle in the same lane.

The whole purpose of double-wide lanes is so that people making right turns don't impede people not making right turns. You don't need that extra space for any other purpose. If you can't go past someone making a right hand turn, then the whole reason for the lane is defeated.

And that ignores the question, did the car not see the bus or did the extra-smart computer just assume that the human would yield, as did the extra-smart human driver of said AV?

Re:Buses have right of way

[perpenso]

I do not believe that most jurisdictions require you to take action to avoid someone else hitting you.

In California you actually are required to do so if it can be done safely.

Re:Buses have right of way

[Anonymous Coward]

The law (no passing in the same lane) is not expected to be followed. It exists so it is possible to pass judgement in a situation like this.

So legally, I would tend to fault the bus. Realistically, I would probably fault the google car.

Re:Buses have right of way

[Anonymous Coward]

I do not believe that most jurisdictions require you to take action to avoid someone else hitting you.

In California you actually are required to do so if it can be done safely.

anyone have a link to the DMV rules about that?

Re:Buses have right of way

[perpenso]

I do not believe that most jurisdictions require you to take action to avoid someone else hitting you.

In California you actually are required to do so if it can be done safely.

anyone have a link to the DMV rules about that?

"Never assume other drivers will give you the right-of-way. Yield your right-of-way when it helps to prevent collisions."
http://test-www.dmv.ca.gov/pub...

My understanding is that this is a paraphrasing of what is in the Motor Vehicle Code.

Re:Buses have right of way

[Anonymous Coward]

Here's a tip that will save you a lot of frustration, expense, and potential injury or loss of life: NEVER insist on having the right of way if doing so will cause an accident.

Oh SHIT, a random internet person says a Google car failed with zero knowledge or experience of what actually happened beyond a poorly written article. It is official people, self driving cars are a failure, scrap them all and cancel the whole idea. It's not 100% perfect from the first second so it can never happen, ever. Delete all research and destroy all materials.

Re:Buses have right of way

[LynnwoodRooster]

the bus was behind and starting to pass the AV on the left side

To hit the side of a bus, it has to already be passing you.

The bus was in the wrong - it was passing another vehicle in the same lane.

The whole purpose of double-wide lanes is so that people making right turns don't impede people not making right turns. You don't need that extra space for any other purpose. If you can't go past someone making a right hand turn, then the whole reason for the lane is defeated.

If the lane is blocked, then you have to go around it. And it's still a single lane - it does NOT have a marked "right turn" lane. One at a time, and passing another vehicle in the same lane is illegal. Even if they're over to the side - it's not legal to pass them. If they come back - you're at fault.

Re:Buses have right of way

[Maxo-Texas]

one thing that will certainly be true is that there will be no "he said/she said" in this case. Google will have a recording of everything about the accident.

If the car was at fault, the software will be tuned.

Accidents happen. So far, they happen less frequently with automated cars.

Re:Buses have right of way

[Jack Griffin]

If the Google car really was traveling at just 2 mph, then you have to wonder whether the bus driver could have avoided the accident.

If the Google car really was traveling at just 2 mph, then you have to wonder whether it could have avoided the accident.

Re:Buses have right of way

[Anonymous Coward]

I do not believe that most jurisdictions require you to take action to avoid someone else hitting you.

Which means: if you're standing still, you can't be blamed for the accident. It has to be that way, as a car can break down anywhere.

On the other hand, "right of way" is not a "right to cause an accident". If someone else breaks the rules when you have right of way, you still have to try to avoid the accident by braking. If you see him, if you have time to react, etc. But "He made an illegal turn so I rammed him" won't fly.

Re:Buses have right of way

[jittles]

I do not believe that most jurisdictions require you to take action to avoid someone else hitting you.

In California you actually are required to do so if it can be done safely.

anyone have a link to the DMV rules about that?

"Never assume other drivers will give you the right-of-way. Yield your right-of-way when it helps to prevent collisions." http://test-www.dmv.ca.gov/pub... My understanding is that this is a paraphrasing of what is in the Motor Vehicle Code.

Yielding right of way is not the same as requiring someone to avoid accidents. Of course you are expected to follow at a safe distance and allow yourself room to stop. This is to prevent rear end collisions. but if someone runs a red light and pulls out in front of you and your options are A) Make an unsafe lane change and B) plow into the guy then everyone will reasonably expect you to do B. Your requirement to avoid an accident is on the basis of you driving in a safe manner and requiring someone to take any other measure besides braking to avoid an accident is dangerous. In the case of yielding right of way, you are slowing or stopping to avoid an accident.

Re:Buses have right of way

[hey!]

If you don't think very hard about it. There isn't much you can do in the way of slowing down from 2 mph, so the only way it could have avoided the accident is speeding up to get out of the way, which might have worked, or might have made the accident worse. We can't be sure.

Re:Buses have right of way

[rhazz]

This is true in Ottawa, however those rights-of-way are very specific and would not apply at all in this scenario. In Ottawa the law is that you have to yield to a bus attempting to merge into your lane from the right side. This is to allow the bus to pull out from a stop after picking up passengers. It is necessary because buses accelerate slowly and on busier roads they need others to yield to let them merge safely. There is a yield sign clearly affixed to the extreme left side of the bus's rear panel.

Re:Buses have right of way

[perpenso]

I do not believe that most jurisdictions require you to take action to avoid someone else hitting you.

In California you actually are required to do so if it can be done safely.

anyone have a link to the DMV rules about that?

"Never assume other drivers will give you the right-of-way. Yield your right-of-way when it helps to prevent collisions." http://test-www.dmv.ca.gov/pub... My understanding is that this is a paraphrasing of what is in the Motor Vehicle Code.

Yielding right of way is not the same as requiring someone to avoid accidents. Of course you are expected to follow at a safe distance and allow yourself room to stop. This is to prevent rear end collisions. but if someone runs a red light and pulls out in front of you and your options are A) Make an unsafe lane change and B) plow into the guy then everyone will reasonably expect you to do B. Your requirement to avoid an accident is on the basis of you driving in a safe manner and requiring someone to take any other measure besides braking to avoid an accident is dangerous. In the case of yielding right of way, you are slowing or stopping to avoid an accident.

The CA DMV manual explicitly says to not bunch up, to leave an open space in one's adjacent lane so that one can make such an emergency maneuver. The manual instructs you to do so if necessary and safe to do so. The DMV instructions with respect to avoiding an accident are not only about yielding.

Re:Buses have right of way

[Anonymous Coward]

The whole purpose of double-wide lanes is so that people making right turns don't impede people not making right turns.

Yes, but it's still only a single lane and still illegal to pass on the left. The purpose of the wider lane is so that you don't have to slow down while the vehicle in front is actually making the turn... you don't get to pass it before it starts to actually enter the turn.

You don't need that extra space for any other purpose.

Except in this case, it DID need that space, to avoid an obstacle in the road.

And that ignores the question, did the car not see the bus or did the extra-smart computer just assume that the human would yield, as did the extra-smart human driver of said AV?

Both the car and the human saw the bus, and determined that since there was plenty of room that the bus would yield as required by law. But the bus driver decided "fuck you, I'm a Bus" and punched the gas, which is why it was already up to 15mph when the google car side-swiped it.

Re:Buses have right of way

[Jack Griffin]

If you don't think very hard about it. There isn't much you can do in the way of slowing down from 2 mph, You can stop can;t you? The stupid google car couldn't even figure this out, hence the crash.

so the only way it could have avoided the accident is speeding up to get out of the way

Also an option, but wouldn't be permissible with AI. Hence we can conclude that the stupid google car is more stupid because it's crash avoidance options are limited

Self-Driving Car with Autonomous Weapons

[fustakrakich]

Just stay out of the way, problem solved

How many autonomous crashes were overridden?

[Your.Master]

I'm a big believer in autonomous cars, but when I see

Google said its car's safety driver thought the bus would yield.

it makes me wonder how many crashes we would have had in autonomous mode, if there weren't an attentive driver who was fully aware he was sitting in an experimental vehicle.

Even if the first rounds of autonomous cars still require a driver for override (for legal reasons if nothing else), it seems like the number of autonomous crashes that likely would have happened is the number has to be driven way down to be comparable to, or less than, the ones with human drivers*; it's not really the amount of autonomous crashes overall that is important.

Also makes me wonder whether any of the manual mode crashes were initiated in autonomous mode and the manual override driver just couldn't recover the situation.

*whether average human drivers or above-average human drivers or even below-average human drivers are the standard is up for debate.

Re:How many autonomous crashes were overridden?

[fisted]

it makes me wonder how many crashes we would have had in autonomous mode, if there weren't an attentive driver who was fully aware he was sitting in an experimental vehicle.

What's your point? Yes, it isn't ready yet. That is why they have a safety driver there in the first place.

Re:How many autonomous crashes were overridden?

[rtb61]

As long as the programmers wear the penalties for what their programs do, including custodial sentences, autonomous vehicles are fine. If they expect the same old, same old, bullshit of you use the program, it's your fault because in the EULA we contrary to all the marketing the program is shite and you are an idiot to use it (marketing says the exact opposite), well, no autonomous vehicles should be banned. Sorry but there is now way in hell I want to share a road with software with the typical EULA no warranty ie this software is worth five dollars, we will only cover you for this value, write of the car become a paraplegic and run down an active school crossing mowing down children needing to be supported for life, it is all your fault for using our shitty software and here's your five dollars back.

So yeah, let's see those warranties first before any more is said about autonomous vehicles.

Re:How many autonomous crashes were overridden?

[ColdSam]

Sure, all of what you say is possible, we don't have all the data. It's ALSO possible that crashes wouldn't have occurred if the self-driving car had been left to correct itself.

However, what data we do have suggests that the combination of AV and scrupulous test driver is better than the average driver. We also know that every time the driver has to take the wheel that it will make the next generation of AV that much better.

Re:How many autonomous crashes were overridden?

[Junta]

Well the point would be that the state of public discourse is based around the assumption that the cars have never caused an accident because they wouldn't. Stories were written all the time about how there were no known instances of an accident where the autonomous system were at fault. That dialog could be disingenuous if the safety drivers intervene often. For example, one accident that was caused by a 'safety driver' was when the car gave up trying to make a left term and the human had to try and messed up. Another (non-google) ride along documented how the autonomous system beeped a warning and disengaged when faced with having to merge with high speed, busy traffic.

That's the side of the story that's not often highlighted, that the parties that be are (rightfully) playing it very safe and not taking a lot of chances that a human driver would have to take on. As a result, a lot of people view this as a utopian super-cruise control that is ready to drive their car wherever, rather than a somewhat controlled bunch of research.

Re:How many autonomous crashes were overridden?

[nukenerd]

it makes me wonder how many crashes we would have had in autonomous mode, if there weren't an attentive driver

What's your point?

The point is that the safety driver's presence and power to intervene means that we cannot rely on the accident rate statistics racked up so far.

Re:How many autonomous crashes were overridden?

[complete loony]

Every time the operator takes the wheel, google use the captured sensor data to write another test case.

Re:How many autonomous crashes were overridden?

[Dutch Gun]

Some companies have already declared that they'll assume liability for their autonomous vehicles. They do this knowing full well that autonomous vehicles are going to be an order of magnitude or two safer than human drivers, mainly because human drivers, on average, are pretty terrible drivers.

Quite frankly, I'm much concerned about sharing the road with other humans who get distracted, don't pay attention, or drive impaired / recklessly around me, and I'm very much looking forward to the day when vehicular-related deaths are a much more rare occurrence than they are today.

BTW, one more thing humans are notoriously bad at is risk assessment.

Re:How many autonomous crashes were overridden?

[rtb61]

Dude, we a talking typical lying jack ass corporations, let's see it in writing because before then it is just empty promises. I did read through the various claims and articles and they are just empty claims subject to government regulatory testing and government vehicle licensing, with a government implemented supportive framework (immediate out, not our fault, the governments fault, never ever forget lobbyists hard at work privatising profits and socialising losses). Here's is betting they will immediately exclude custodial sentences, I kill someone through negligence and I go to jail, so the programming team kills someone through negligence and they also should go to jail, now that is full liability and I'll bet every cent I have and many times that, that Volvo is guaranteed to exclude that. So they lied, straight up lied from the get go, "Volvo will accept full liability", is a straight up lie because they will deny "CRIMINAL LIABILITY".

Re:How many autonomous crashes were overridden?

[Anonymous Coward]

Even if the first rounds of autonomous cars still require a driver for override (for legal reasons if nothing else), it seems like the number of autonomous crashes that likely would have happened is the number has to be driven way down to be comparable to, or less than, the ones with human drivers*

Car has normal everyday driver. Driver is not driving, but monitoring. Driver is not driving but mentally planning... Driver is not driving, but browsing various web sites. Drive is not driving, but watching Robocop... In short, I just can't see the average driver really watching all that closely, at least once they are comfortable with it. After that you have

1) Car doesn't warn at all. Crash.
2) Car warns driver too often of possible conditions. Again, attention may not return quickly enough as it will be assumed to be a false alarm. Crash.
3) Car warns too seldom. Incidents are missed anyway. Crash.
4) Car warns just right. Is there such a thing? Even then, the human has to task switch. Time is up. Crash.

Crashes might still be much less, but I'm not convinced that a human driver ready to override will perhaps do as much as advertised. Better than nothing maybe, but most likely if the person isn't paying close attention, which he probably won't be once the car drives itself, then the bad things will be missed till it is too late, the same way they are when people text/etc. Also, if the car's system knows there is something wrong, it likely is already taking action and not waiting on the flesh unit to do something, so basically most of 2-4 are probably handled already (or will be), and it might be more risky to second guess the car, particularly if you weren't paying close attention.

Actually I can see one area where the human might help. If the car warned you that road conditions were hazardous because of snow, storm, wind, lane closures, etc. Then you might get the human to pay enough attention to matter. A car's computer can internally estimate just how close it is from losing traction. That information could be passed between cars, and help in extreme conditions to correct the internal systems estimates as to the max safe velocity.

Re:How many autonomous crashes were overridden?

[Wootery]

You place undue emphasis on the programmers.

If the programmers are working in an environment with poor quality-control, and they're being told to implement a bad spec, would you blame the programmers?

Re:How many autonomous crashes were overridden?

[Anonymous Coward]

You said it yourself. "Experimental".

Re: How many autonomous crashes were overridden?

[Anonymous Coward]

Do you really want software developers here (as a class?) to assume more liability than any licensed Professional Engineer does? Or does Crashy Motors, Inc. appoint one scapegoat for all their cars?

Re:How many autonomous crashes were overridden?

[Anonymous Coward]

we cannot rely on the accident rate statistics

i.e. we need more information. we need to know how many times the driver intervened and to what extent (turn slightly vs sharply; brake lightly vs abrupt; signal; etc)

Re:How many autonomous crashes were overridden?

[Anonymous Coward]

Seems like it wasn't in autonomous mode if the safety driver was actively deciding what the car should do

Re:How many autonomous crashes were overridden?

[virtual_mps]

I'm a big believer in autonomous cars, but when I see

Google said its car's safety driver thought the bus would yield.

it makes me wonder how many crashes we would have had in autonomous mode, if there weren't an attentive driver who was fully aware he was sitting in an experimental vehicle.

Probably 1. The research suggests that having a human doing nothing except watching and waiting to override a failing autonomous system basically never works--so it probably isn't a big factor in how many accidents google has.

Re:How many autonomous crashes were overridden?

[ColdSam]

Probably far more than that, actually. First, these operators were far more attentive and knowledgeable than the average expected lump behind the wheel and they were on the job, expected to take over (and likely losing their job if they were goofing around). Second, they would likely preemptively take the AVs out of autonomous mode in situations that they were unlikely to be able to handle well.

Re:How many autonomous crashes were overridden?

[rtb61]

As this is about cars and using the car analogy, would you blame the driver of a vehicle for killing a family member, when the driver knew the vehicle was shoddy and would not function properly and they would likely run into people and injure them but got into the car and drove it anyhow, YES! So the programmer is the driver and the program fails and as Volvo has claimed they will take full liability and that would include liability for criminal negligence, then yes, the coders goes to jail, the supervisor goes to jail and the administrator goes to jail, in autonomous vehicles they take on the role and responsibilities of the driver.

Re:How many autonomous crashes were overridden?

[cwsumner]

If the Engineers are working on a bridge in an environment with poor quality-control, and they're being told to implement a bad spec, would you blame the Engineers?

The answer is Yes, but not just them, others as well.

Re:How many autonomous crashes were overridden?

[ColdSam]

This applies only if the programmers knew that the work was substandard and likely to cause unnecessary accidents. The analogy would be closer to a manufacturer and factory workers knowingly building and installing faulty parts (e.g. airbags). There would be no criminal liability in either case if the workers did not know and could not be expected to know the faults.

Can you smell the lawsuits?

[WillAffleckUW]

I know I can.

No jury would find against the human and in favor of the robot.

Re:Can you smell the lawsuits?

[Anonymous Coward]

I would, the robot bus driver would have slowed and avoid the accident instead of being a grumpy asshole that insisted on taking his right of way.

When they're all robots, the cars would have been in communication rather than needed to predict anything. The sooner humans are removed from the controls of vehicles the better off we'll all be.

Re:Can you smell the lawsuits?

[__aaclcg7560]

When they're all robots, the cars would have been in communication rather than needed to predict anything.

Unless the car is infected with the A-HOLE virus, then it takes two parking spots for parking. Something that Hummer owners used to do at a job I've worked at.

Re:Can you smell the lawsuits?

[Jeremi]

No jury would find against the human and in favor of the robot.

I would find against the human, if the evidence showed that the human were in fact at fault. I have enough faith in the rationality of my fellow man to predict that most jurors would do the same.

And with a self-driving car there will be no lack of evidence regarding exactly what happened, since it keeps a full audit trail of everything that occurred before and during the accident. (Compare with a typical accident involving human drivers, which often devolves into a he-said-vs-he-said situation, with both sides offering only a partial and subjective subset of the truth)

Re:Can you smell the lawsuits?

[WillAffleckUW]

I've served on juries.

I stand by my statement.

Re:Can you smell the lawsuits?

[Obfuscant]

I would find against the human, if the evidence showed that the human were in fact at fault.

Have you been on a bus that suddenly slammed on the brakes because a car just pulled out in front of it? And you'd find the bus driver at fault for not causing injuries to the bus passengers?

The AV was going two miles per hour. That means it was just starting up. The bus was going 15. That means it had been in motion for a much longer period of time. The amazing perfect computer car couldn't see the bus before pulling over, or did it just not care?

Re:Can you smell the lawsuits?

[tlambert]

I know I can.

No jury would find against the human and in favor of the robot.

Just because the robot took your job doesn't mean the robot gets a paycheck. Sue the robot all you want: it's not going to be able to pay you.

Re:Can you smell the lawsuits?

[William Baric]

Do you know how many times I had to avoid a bus who suddenly decided to go in my lane? Bus drivers are by far the worse drivers on the road.

Re:Can you smell the lawsuits?

[Jack Griffin]

I know I can.

No jury would find against the human and in favor of the robot.

A robot programmed by a human.
See millions of cases of lawsuits that involve machines, the designer is absolutely liable for any faults.

Dear Google...

[__aaclcg7560]

Neither school buses nor soccer moms ever yield in traffic.

Re:Dear Google...

[RandomFactor]

Change that to TRANSIT buses and soccer moms and i'm with you in Atlanta.

OTOH the school buses in my burb are actually pretty considerate.

Re:Dear Google...

[Anonymous Coward]

OTOH the school buses in my burb are actually pretty considerate.

School buses kind of have to. If the school bus has right of way and crashes into some idiot-driven car- the driver will still get in trouble for the kids that get hurt. It is better to brake carefully - getting a little late is better than catching hell & lawsuits from hysteric parents. And nobody hires the school bus driver with a bus crash on his resume.

A MUNI Bus yield?

[Anonymous Coward]

That car must be drunk

thought the bus would yield

[avandesande]

1st rule of defensive driving- never expect another driver to do anything

I won't yield into traffic or turn into a street if another driver will need to slow or brake not to hit me
Never sit in a median to turn with front or back of car sticking out
I actually speed up a bit before turning to maximize distance between myself and driver behind and turn shallow. This is a bit hard to explain but you angle into the turn and actually do most of your slowing when you are already in the turn
Many others but I probably don't even think about them.

Re:thought the bus would yield

[Anonymous Coward]

I actually speed up a bit before turning to maximize distance between myself and driver behind and turn shallow. This is a bit hard to explain but you angle into the turn and actually do most of your slowing when you are already in the turn

Sounds very risky, but that might be because I live in a place where the roads usually are covered with either rain or ice.

Re:thought the bus would yield

[Anonymous Coward]

It's only risky if driver is going fast enough for it to matter. Something tells me this guy sticks to the speed limit. And maybe that Google car would have whipped by him at 2 mph. My grandmother would probably yell at him to get out of the fast lane. But yeah, driving 101 says that you don't brake during a turn because you lose grip, and hence control.

Re:thought the bus would yield

[jittles]

1st rule of defensive driving- never expect another driver to do anything I won't yield into traffic or turn into a street if another driver will need to slow or brake not to hit me Never sit in a median to turn with front or back of car sticking out I actually speed up a bit before turning to maximize distance between myself and driver behind and turn shallow. This is a bit hard to explain but you angle into the turn and actually do most of your slowing when you are already in the turn Many others but I probably don't even think about them.

You should probably take a driving safety course. Speeding up or slowing down in a turn requires traction. Your traction in a turn is always a fixed amount (that varies on conditions). With 4 wheels, this may not be a huge problem in favorable conditions. With two wheels, this can be very dangerous. I hate when people do exactly what you describe while I'm on my motorcycle. I do not want to touch my brakes in a turn unless its an emergency. I try to maintain a constant speed. Even if I coast to slow down, I slow down much more slowly than a car when turning.

Re:thought the bus would yield

[Ichijo]

I actually speed up a bit before turning to maximize distance between myself and driver behind and turn shallow. This is a bit hard to explain but you angle into the turn and actually do most of your slowing when you are already in the turn

That's illegal in California. You're supposed to make the turn from as close as practicable to the right-hand curb or edge of the roadway.

Re:thought the bus would yield

[Anonymous Coward]

Do that in an old Porsche 911 on a slippery surface and you will soon see the error of your ways.

Ask me how I know...

AC

Re:thought the bus would yield

[Anonymous Coward]

1st rule of defensive driving- never expect another driver to do anything

This is why I get a little prickly by reports that this would be the first time Google's Autonomous Car might have "caused" something. It not being at legal fault is not the same as it not causing something. If a human could have avoided a collision that would have not been their fault, then the autonomous car caused the crash.

I.e., the real test isn't whether or not Google is at fault less often than a human, it's whether or not the total number of collisions is decreased using the autonomous car versus not. If you randomly assigned a population Google cars, and another comparable population the same car with out AI, would you see an increase, decrease, or the same number of accidents?

I don't give a @#$*! who is at legal fault. If I die but it's the other's fault, I still lose.

Re:thought the bus would yield

[Anonymous Coward]

I actually speed up a bit before turning to maximize distance between myself and driver behind and turn shallow. This is a bit hard to explain but you angle into the turn and actually do most of your slowing when you are already in the turn

This is a terrific way to lose traction and spin out, and I don't see what the benefit is supposed to be. How is this "defensive"?

Slowing during the turn implies loss of traction. And if you're braking (are you? Or are you just coasting down in speed?) you're losing a hell of a lot of traction. Brake before the turn, use the throttle carefully to maintain speed until you're through, then gently accelerate out.

Re:thought the bus would yield

[Anonymous Coward]

Actually, the first rule of defensive driving is to expect another drive to do everything. IOW, defensive driving is about
1) What should I do if the driver stops
2) What should I do if the driver doesn't stop
3) What should I do if the driver turns left
4) What should I do if the drive turns right ...
Profit.

Re:thought the bus would yield

[HornWumpus]

For front drive: You'll never get you inside rear wheel off the ground driving like that.

For rear drive: No mention of opposite lock or throttle steer?

Weight transfer. Look it up. It's not so simple as you say.

Re:thought the bus would yield

[Anonymous Coward]

I would honestly recommend that you instead expect the other driver to do the worst possible thing, rather than not expect them to do anything.

In the case of you speeding up in a turn to create distance, say I speed up to match? Now you're in a worse predicament.. and so am I when you slow down. Better to slow down slightly in advance and expect me to do the same. If I don't it's not your fault and since we were going slower the damage is less. Plus you don't have to lose control or hit something you didn't expect when going around that corner of yours..

Re:thought the bus would yield

[Anonymous Coward]

1st rule of defensive driving- never expect another driver to do anything

I won't yield into traffic or turn into a street if another driver will need to slow or brake not to hit me

Never sit in a median to turn with front or back of car sticking out

I actually speed up a bit before turning to maximize distance between myself and driver behind and turn shallow. This is a bit hard to explain but you angle into the turn and actually do most of your slowing when you are already in the turn

Many others but I probably don't even think about them.

You should probably take a driving safety course. Speeding up or slowing down in a turn requires traction. Your traction in a turn is always a fixed amount (that varies on conditions). With 4 wheels, this may not be a huge problem in favorable conditions. With two wheels, this can be very dangerous. I hate when people do exactly what you describe while I'm on my motorcycle. I do not want to touch my brakes in a turn unless its an emergency. I try to maintain a constant speed. Even if I coast to slow down, I slow down much more slowly than a car when turning.

well if you really were a cautious motorcycle driver, you wouldn't be so close behind avandesande for him to try and make room by speeding up before the turn

Re:thought the bus would yield

[jopsen]

1st rule of defensive driving- never expect another driver to do anything

2nd rule: always expect another driver to do anything :)

Particularly in America, people completely ignore all sanity here, be it stop signs (even without any oversight), blinking, illegal u-turns the list goes on...

Re:thought the bus would yield

[Jack Griffin]

1st rule of defensive driving- never expect another driver to do anything

Try riding a small underpowered motorcycle. It will teach new a whole new appreciation for defensive driving.

Re:thought the bus would yield

[Anonymous Coward]

1st rule of defensive driving- never expect another driver to do anything

That's a nice thought, but if you actually attempt to follow that to an extreme you'll never actually turn the vehicle on to start with. That rule is talking about things like don't assume that a turn signal means a car is going to turn- you also need to consider the speed of the vehicle, whether it's slowing down or not, and look at the driver for visual cues as well. It means that you shouldn't assume the car in the lane next to you will not suddenly swerve into your vehicle, but rather 'keep an eye' on it and make sure it's not suddenly moving towards you. Etc.

This is a bit hard to explain but you angle into the turn and actually do most of your slowing when you are already in the turn

... then the first time you hit loose gravel, water, ice, etc. or are going too fast, you're going to throw yourself into a skid or spin, hurtling your vehicle across the road into oncoming traffic. You brake BEFORE you enter the turn, and then accelerate THROUGH the turn. This is not only basic physics, but also a part of any worthwhile driving instruction.

I won't yield into traffic or turn into a street if another driver will need to slow or brake not to hit me

What the hell does "yield into traffic" even mean? And unless you're the only one on the road, any time you slow down for any reason, the guy behind you will have to brake, so that doesn't make any damn sense either.

You seem to be confusing "defensive driving" with "timid driving", which is actually the opposite of defensive driving.

Re:thought the bus would yield

[avandesande]

Thank you for your idiotic suggestion. I am not doing a road rally, I am probably at 1/3 of the road engineering speed and the risk of getting hit in the ass is probably a 1000 times greater than skidding.

Re:thought the bus would yield

[jittles]

Thank you for your idiotic suggestion. I am not doing a road rally, I am probably at 1/3 of the road engineering speed and the risk of getting hit in the ass is probably a 1000 times greater than skidding.

My idiotic suggestion of slowing down prior to entering the turn? If you get rear ended with a normal braking procedure then the person behind you is an idiot. There's no rally racing about it. You want to pick your speed prior to entering the turn, maintain that speed during the turn, and accelerate coming out of the turn. It provides the best traction possible no matter what kind of vehicle you're in. If you're slowing down in the turn then you are wasting traction. If you're going 1/3 of the speed that the turn is designed for then why are you slowing down at all?

Re:thought the bus would yield

[Anonymous Coward]

What's the definition of "practicable"? Possible to do in a safe manner? It could be argued that minimizing the turning angle reduces the combined slowing before and traction during turn requirements and is therefore an increase in safety.

I think it's also important to make a distinction between a turn (as in leaving the current road) and a curve (current road moves in a new direction). The OP is still wrong, it's best to slow well in advance or otherwise signal to the following driver to allow them time to adjust following distance as is optimal before sudden braking. On the other hand, there isn't usually a signal required to following drivers in the case of a curve, since most drivers don't assume you're just going to keep driving at max speed and fly into the ditch.

Re:thought the bus would yield

[Anonymous Coward]

I hope you and every other motorcyclist and cyclist dies in a painful accident soon, so the roads will be safe for two-legged humans and four-wheeled vehicles. I am sick of taking special precautions to avoid being injured by or injuring you spoiled recreational garbage.

Look what's next on my feed

[Anonymous Coward]

On my news feed, a couple lines below it I have the following headline:
Jalopnik Idiot Street Racers Blamed For Massive Freeway Wreck That Killed Three.

I can't wait to see google cars drag racing.

Talented Slashdotters

[avandesande]

Could somebody please come up with a fitting car analogy?

Re:Talented Slashdotters

[Anonymous Coward]

Group using dial-up might have overloaded netflix servers

Re:Talented Slashdotters

[Anonymous Coward]

Google car sandbags, then catches the bus.

Re:Talented Slashdotters

[Anonymous Coward]

Sorry, can't help you.

BUT

"Fitting car analogy" is an anagram for "a falcon gyrating it", if that's interesting to you at all.

Google is now taking some responsibility

[Lucas123]

Google is now saying they were following the "spirit of the road" when the crash happened and that they've reviewed the incident, as well as thousands of variations on it, in a driving simulator and made refinements to its AV software.

Re:Google is now taking some responsibility

[Anonymous Coward]

And there in lies the rub. No Google car will make the same mistake again. Likely no other autonomous car will make the same mistake again. And thus by having a minor fender bender during beta testing we prevent hundreds of collisions in the future. No matter how many times a human did the same thing, more humans would continue to make the same mistake, until we could come up with a law to prevent it. i.e. always yield to buses no matter what.

Re:Google is now taking some responsibility

[Anonymous Coward]

And that is how we should expect every autonomous car maker to own up to a claim.

Re:Google is now taking some responsibility

[Anonymous Coward]

Except for the fact that "yield to buses no matter what" may not always be the safest or most prudent decision to make. And there in lies the rub, machines can only do what they're programmed to do there are so many infinite number of situations that they will always be effectively useless except for driving on controlled-access highways.

Re:Google is now taking some responsibility

[Anonymous Coward]

No Google car will make the same mistake again.

Have you read the linked article? It even spells it out that they only changed a variable for "large" vehicles - the next crash could involve the same algorithm, the same decision and a family car, followed by a motorcycle and lastly an old man crossing the street.

Re:Google is now taking some responsibility

[Anonymous Coward]

more humans would continue to make the same mistake, until we could come up with a law to prevent it.

As if that would solve the problem? Have you never gotten a ticket?

Re:Google is now taking some responsibility

[Anonymous Coward]

No matter how many times a human did the same thing, more humans would continue to make the same mistake, until we could come up with a law to prevent it. i.e. always yield to buses no matter what.

And then there would still be idiots breaking that law.

Re:Google is now taking some responsibility

[Anonymous Coward]

Y'know, I'd actually pay to see an old man crossing the street sideswipe a Google car while going 15 mph in the same lane...

Re:Google is now taking some responsibility

[Anonymous Coward]

Until we can come up with a law to prevent it? Humans cannot be reliably expected to follow the law, hence why traffic tickets and police officers exist.

That mistake will continue to be made by humans for all eternity.

Re:Google is now taking some responsibility

[houghi]

No Google car will make the same mistake again. Likely no other autonomous car will make the same mistake again.

Why no other one? Is this an open source project, or are they sharing, like Volvo who said safety is more impotant than having the patent on the 3-point safetybelt.

It would be interesting if the companies (willing or unwilling) would share this kind of data. I am sure hat that will not happen. It will be 'good that their car killed a person, that way we sell more.'
Because you know that at one point it WILL kill somebody, even if it is to avoid killing 2.

Re:Google is now taking some responsibility

[cwsumner]

And there in lies the rub. No Google car will make the same mistake again. ...

Don't be silly, of course they will. Google's fixes are no more perfect than anything else.

Re:Google is now taking some responsibility

[Anonymous Coward]

...implying that humans know about the laws and will be able to think of them when the time comes

Google Clown Car

[Anonymous Coward]

Maybe Google should send the car with Cheech and Chong!

Ha ha

Google Car Programming Missed a Law

[Anonymous Coward]

Newton's Right of Way Law: the biggest one has the right of way. In most large cities, that supersedes the Vehicle Code. Buses have the right of way if they want it. Taxis take the right of way. Uber ignores the right of way (stop in front of you without warning to pick up a passenger?). Prii and smaller (alias road boulder) must grant r/w to anything else that wants it.

Car analogy

[Anonymous Coward]

This is just like when you have a self driving car that is trying to avoid sandbags and hits a bus instead.

In the same lane?

[singularity]

So based on numerous descriptions I have read, the Google car was in a very wide lane and moved to the right side of the lane to make a right turn. It saw some sandbags blocking the very right side of the lane, so it tried to move back to the middle of the lane. A bus, coming up from behind in the same lane, did not yield to to the Google car and there was contact.

I think it is important to note that all of this happened in the same "lane".

While the Google car could have possibly avoided the accident, I am not sure it is to blame. It seems to me that the bus was attempting to pass a car ahead of it in the same lane.
The blame seems about 80% on the city for not properly marking the lanes, about 15% on the bus for not yielding to a car ahead of it in its own lane, and about 5% on the Google car for not stopping for the bus who was trying to barge its way through.

Re:In the same lane?

[houghi]

So you read an article online. and while the Google people, who have access to all data, who have examined it in various ways, who have altered the code, because they say the car was in the wrong. THEY SAID THE CAR WAS IN THE WRONG, you can come up with statistics that show that the blame is not on the car?

So in effect you disagree with the Google people who have the actual data and alter it with you made up data.

It's up to the level of human drivers

[SuperKendall]

The accident was because the car saw sandbags on the right and in an overabundance of caution decided to move a whole lane over, into a bus.

Well that's as good as many human drivers who I have seen swerve from the lane they are in at seemingly nothing without a glance, and absolutely why you do not linger in someones blind spot.

An open question though is how it saw sandbags and not a BUS...

Re:It's up to the level of human drivers

[Jack Griffin]

Well that's as good as many human drivers ...

Many but not all. The point for me is that it is less good than myself, so why would I buy one?

Re:It's up to the level of human drivers

[rl117]

The other question I have from this, is why it was moved all the way over into the right lane if it saw a sandbag there?

If it was me, I'd have kept centred in the lane until after the sandbag, then moved fully over once past it. That would preclude any risky overtaking by any traffic wanting to go in the left lane, and any side collisions since I'm changing lane only once rather than effectively three times. Is the google algorithm a bit over-eager on lane changes?

Re:It's up to the level of human drivers

[ColdSam]

The obvious answer is that it didn't see the sandbag there until it had moved over.

Not the first time

[NicknameUnavailable]

The "safety driver" has just been overriding all other would-be accidents.

Buses YIELD?

[superdave80]

Google said its car's safety driver thought the bus would yield.

BWAHAHAHAHA!!! Has this guy ever driven in SF before? A bus YIELDING to another car? In your dreams. I drove through SF for years, and buses didn't give a crap who was around them. When they pulled off to pick-up/drop-off passengers, they would intentionally park at an angle to keep the lane blocked so they could more easily start back into the lane. Even if they didn't block the lane, if they wanted to get moving in that lane, they just go. They know that they are bigger than the cars, so they know the car will slow down or move out of the way. If a lane became more narrow than they liked due to parked cars on the side of the road, they would just take up two lanes. If you are next to it in the lane that they now want to occupy? You better move the fuck over. They would run red lights at will. Watch out if you are trying to cross at an intersection with a traffic light and a bus is coming through. Without a doubt, bus drivers in SF were the worst.

Re:Buses YIELD?

[Anonymous Coward]

You talk like you don't know the traffic laws in most areas. Buses don't yield because they have the right of way. You are supposed to yield, plain and simple.

Re:Buses YIELD?

[Anonymous Coward]

And this is what will have to change. With technology, surveillance and other measures, these drivers will be made accountable for their irresponsible driving. Eventually it will become a thing of the past (due to hefty penalties being the consequence). Autonomous cars are the future, and primitive drivers will be forced to adapt.

Re:Buses YIELD?

[Anonymous Coward]

I've been on the bus in SF and I currently reside in a very suburban city with poor transportation. I have to say I prefered the aggressive form of presence that public transport has there. Not in just a literal sense but in terms of driving tactics and mentality as you mentioned above. I was on the bus, chatting with the bus driver, and we actually counted two minutes before any of the fucking drivers (in the shockingly more suburban part of town) would let us through. We actually had to wait until there was no one there. I think car drivers everywhere need to give more respect to public transportation and pedestrians.

Driving with a very yielding and respectful style/mentality is respectful to your fellow citizen. And no, "at least I don't run you over right?" doesn't count as respectful..

Re:Buses YIELD?

[Anonymous Coward]

The car wasn't in SF...

Re:Buses YIELD?

[Anonymous Coward]

More like "has this programmer ever driven anywhere at ALL?" Never expect the larger vehicle to yield, it's simply a matter of physics. Only dumbasses cut off larger vehicles.

Re:Buses YIELD?

[Anonymous Coward]

Sounds like a good call for some autonomous buses in SF.

Google's cars made illegal right turns all along

[Ichijo]

Most of the time Google's AVs drive in the middle of a lane but "when you're teeing up a right-hand turn in a lane wide enough to handle two streams of traffic, annoyed traffic stacks up behind you.

"So several weeks ago we began giving the self-driving car the capabilities it needs to do what human drivers do: hug the rightmost side of the lane.

The law says a right-hand turn shall be made as close as practicable to the right-hand curb or edge of the roadway. So Google's self-driving cars have been making their right turns illegally until just recently.

I expected better from Google.

When they find who the bus driver was..

[Anonymous Coward]

When they find who the bus driver was, they will be blocked from Google search for not yielding to the Google car.

'Caused'

[Martin S.]

Agreed, reading the report it is very obvious the bus actually caused the accident by trying an inappropriate overtake the Lexus

The Lexus only 'Caused' the accident only in so much as it did not avoid it.

can't wait for the self driving car bubble to pop

[Anonymous Coward]

...so that resources can be put to better use in driving assistance, rather than autonomous driving, which is a dangerous nonsense. Real life situations are far too complex for algorithms to evaluate and react to, on tiny roads full of obstacles and uncontrollable events.
The only way to fully automate cars would be to have separate roads, with predetermined paths either physically determined or via f.e. radio signals, adapting technology and experience from air travel like cruising autopilot and ILS.
Forcing autonomous vehicles to adapt to an analogue path instead of creating a digital path for the vehicles to follow is just approaching the problem from a hopeless angle.

Safety driver doesn't know rules of the road

[Anonymous Coward]

The bus has the right of way. Google's car should have yielded, plain and simple.

Track record

[Anonymous Coward]

Our car had no accidents in autonomous mode because it'll be yanked into manual mode if the safety driver percieves danger.

What's not clear...

[Anonymous Coward]

...is whether or not it was possible to drive over the sandbags.

A car with no driver moving at 2mph

[h33t l4x0r]

Probably looks like someone forgot to set their parking break.

"Where do you draw the line?"

[tlambert]

Where do you draw the line?

Typically, it is drawn between the lanes...

Or one court case...

[tlambert]

That's proprietary info, you are not going to have access to it until after a couple of gigantic court cases involving dead children.

Or one court case involving a gigantic dead child?

Re:Or one court case...

[Anonymous Coward]

Please leave the slashdotters out of this.

Expected the other guy to yield ...

[perpenso]

How can you possibly determine that someone willfully failed to avoid the accident?

When one side says "Google said its car's safety driver thought the bus would yield." When a driver sees a potential accident coming the proper response is not to bet the other guy will yield, even if you have the right of way. Sometimes you have to accommodate the other guys illegal move.

Re:Expected the other guy to yield ...

[jittles]

How can you possibly determine that someone willfully failed to avoid the accident?

When one side says "Google said its car's safety driver thought the bus would yield." When a driver sees a potential accident coming the proper response is not to bet the other guy will yield, even if you have the right of way. Sometimes you have to accommodate the other guys illegal move.

Sure, the guy admitted that he made a mistake. I feel like there are other laws that cover this scenario, though.

Re:Expected the other guy to yield ...

[tehcyder]

How can you possibly determine that someone willfully failed to avoid the accident?

When one side says "Google said its car's safety driver thought the bus would yield." When a driver sees a potential accident coming the proper response is not to bet the other guy will yield, even if you have the right of way. Sometimes you have to accommodate the other guys illegal move.

Call me a wuss, but I work on the assumption that it's always safer to give the benefit of the doubt to the much larger vehicle.

Re:Expected the other guy to yield ...

[ColdSam]

"Sees a potential accident coming" is such a broad statement that it is useless. Expecting a car to yield when it SHOULD yield and continuing on does not automatically make you at fault in an accident.

Re:Expected the other guy to yield ...

[perpenso]

How can you possibly determine that someone willfully failed to avoid the accident?

When one side says "Google said its car's safety driver thought the bus would yield." When a driver sees a potential accident coming the proper response is not to bet the other guy will yield, even if you have the right of way. Sometimes you have to accommodate the other guys illegal move.

Sure, the guy admitted that he made a mistake. I feel like there are other laws that cover this scenario, though.

I believe in California the law requires one to yield to avoid an accident even if you have the right of way.

Re:Expected the other guy to yield ...

[perpenso]

"Sees a potential accident coming" is such a broad statement that it is useless. Expecting a car to yield when it SHOULD yield and continuing on does not automatically make you at fault in an accident.

If it is apparent the other vehicle is not yielding despite a legal requirement to do so, you are in California partly responsible if your press your right of way and a collision occurs. And yes it is a broad and general requirement of drivers.

Re:Expected the other guy to yield ...

[ColdSam]

There is a big difference between "may not yield" and "appears to not be yielding". Nobody is saying that you should proceed blindly when an accident is imminent, but to take action whenever an accident is even possible leads to a degenerate and even more dangerous system. The classic case is the freeway onramp where the vehicle on the freeway is rapidly changing speeds to try to avoid the car entering the roadway (who instead of smoothly moving ahead or behind now has to quickly figure out what the other vehicle is trying to do).

This touches on what I said before.

[morethanapapercert]

In previous posts about autonomous cars, I raised the question of how these vehicles handle the highly variable and difficult to anticipate changes in the routing caused by construction. I worked for several years in road construction and can tell you that an appalling number of humans get confused by having to change lanes in response to a flagman or pylons/barrels, ignoring any existing lane, curb and signed markings.

In this case; having read the article (I know, I know...) it seems that the car programming is overly optimistic about predicting the behaviour of vehicles overtaking it. It seems possible that the programming includes implicit assumptions of the likely stopping distance and reaction times it should expect from other vehicles as well. In other words; it "thought" it had sufficient space and time to perform the manoeuvre because it "assumed" a bus would behave and react the way a car might.

I have two thoughts, each in defence of one of the vehicles in this collision:

1) Even the safety driver expected the bus to yield and from I can glean from the article, legally the bus should have yielded. So this was a mistake that even the majority of human drivers might have made in the same situation.

2) Others in this thread have posted criticisms of bus drivers in their city or in general. Much of the annoying behaviours they mention though are pretty understandable from the bus drivers POV. You can't just suddenly hit the brakes if a smaller vehicle or pedestrian darts in front of you. Not only do you have a hell of a lot of momentum (highly variable, depending on passenger load) you also have to make as gradual velocity changes as you can. Your passengers aren't buckled up, you might have a fair number of them standing, with any number of knapsacks, briefcases, skateboards etc etc that become flying hazards when you come stop too suddenly. You have to ease to the left a fair bit when making a right turn because you have a much larger turning radius than most other vehicles. You have to drive straddling lines sometimes because if you stayed tight to the right, you are going to crunch someone, hop the curb or both. On the other hand, if you do stick to the left as much as you can, lots of people are going to pull what Torontonians call a "cabby pass" where the cab illegally passes a bus or streetcar on the right so as to get out from behind it. If they don't use their rear end to block the traffic lane, quite often they'll never get back out because no one wants to stop at the buses back corner and let the bus back in. (I have a relative who is a TTC bus driver and he has passed along some training and daily work anecdotes)

Violated the First Law

[perry64]

Not of robotics, but driving.

I grew up and learned to drive in SF and Rule 1 was always: "Don't mess with MUNI." The quality of their drivers, the driver's attitude ("I get to go home the rest of the day if I hit something?"), as well as the law of gross tonnage made it wise to give them as wide a berth as possible.

Corollary "A" to this rule is "Never expect MUNI to do the sane thing." I would have expected this rule to have been as ingrained in Google's driverless cars as strongly as the first law of robotics was ingrained in Asimov's robots, but I guess not.

Never gonna happen

[Anonymous Coward]

Love technology. Keep hearing people comparing how elevators were once manual now automated.

But Cars are weapons in hands of unpreditables. Too many variables for software to catch up.

Don't let the perfect become the enemy of the good

[GPS Pilot]

I think a 98% reduction in accident rates is pretty easily doable by first-generation autonomous cars. But you know there are some people who don't think in those terms. "Look," they'll say, "here's one that drove into a bus; we mustn't let these things on the road!"

So what is needed to keep this lifesaving technology from being derailed is a concerted effort to educate people that the perfect must not become the enemy of the extraordinarily good.

Google in the wrong.

[TheRealHocusLocus]

Thanks for the street view and concise list! But that double-wide does not look like two lanes, it is a lane and a shoulder for parking. Our city does this too, and people do use it like the Google car intended to but it is not marked for turning. Practically speaking you have to consider the shoulder area near the intersection as a no-mans-land.

According to the report, they were side-by-side in the same, double-wide lane, hence the shared responsibility. Here's a Street View picture of the turn in question [google.com]. Apparently, the sequence was:
1) Red light.
2) Google car signals for a right turn.
3) Google car gets into right side of the double-wide lane and passes cars that are stopped for the red light.

My take is, even though it intended to turn right, the Google car was in the wrong to drift to the right away from the dashed white line it was obviously following on its left side. It was interpreting the space to the right as driveable. For it to be a driveable lane there would have to be either a dashed line to the right that Google car would cross over, OR a right arrow painted on the road ahead indicating it was indeed a lane dedicated to right turns. These things were not there. Google should have stayed in the lane, not have interpreted the right side as a valid to drive at all --- unless it was pulling over to stop or parking. Google car was in the wrong for passing cars to the left of it. It was driving on the 'shoulder', not within its own white-marked lane or designated right turn lane.

4) Google car has to stop because there are sandbags blocking the storm drain.
5) Light turns green, cars start moving.
6) Google car waits for cars to pass to create an opening, then slowly moves back towards the center of the lane.

Google car is in the wrong again because it was stopped and is now moving slowly on the 'shoulder' and is NOT signalling left to indicate it wishes to re-enter the lane.

7) Bus decides not to yield to the Google car that's ahead of it in the lane, trying to pass it anyway.

Google car was probably in 2mph 'crawl mode' which is not human to do at all. Humans either wait or gas 'n go. From the bus driver's relative speed the Google car probably looked like it was stopped. The Google car may even still be signalling right at this point. The bus does not see the sandbags, all it sees is a car on the shoulder.

8) Bus gets its nose a bit ahead of the Google car.
9) Google car doesn't turn the wheel back in time and scrapes the side of the bus.

Google car completely at fault. It was on the 'shoulder' not in the lane. Vehicles ahead of you lose their right of way when they leave the lane. Google car (likely) NOT signaling left to indicate re-entering the lane (because it thought was in a separate and valid lane: wrong). But even if you are signaling left to indicate you wish to enter a lane, you do not gain right of way.

So unless it was exceeding the speed limit the bus is in the right, There is the other thing, that contact was made after the front of the bus passed, which alleviates any suspicion that the bus hit something directly in its path. Google in the wrong, and they need to review the criteria for drivable lanes. If the road has clearly visible lines to your left and there are none to your right, that's NOT a drivable lane over there. Stay in your lane-position, don't pass anyone and (of course) watch your right-rear for dipshits driving on the shoulder.

Something like this happened to me. I'm centered in a marked right turn lane at a solid-circular red light. Cars on the opposite side of the intersection had all turned left and passed in front of me and now no one is coming. It is now clear for me to ta

Re: Google in the wrong.

[Anonymous Coward]

The bus driver failed to have their vehicle under control. Sand bags, ice, glass, flooding... Circumstances dictate safe driving conditions.

Re:Google's cars made illegal right turns all alon

[Anonymous Coward]

I expected better from Google

Why? These are still experimental vehicles, it does not make sense that they would be fully competent at all intricacies of both practical driving and legal driving (as these are not the same things, with moderate overlap) at the experimental stage.

Sticking to the middle of the lane and having "annoyed traffic" stack up behind you is not a terrible state of affairs during this stage, and it better than attempting to have the vehicle stick to the curb when the AI is not ready for it and potentially causing an accident involving pedestrians.

Who will be the first to die by it?

[BrendaEM]

It never had to take a driving test like you did.
It will come defended by one of he largest companies the world has ever seen.
It will put thousands of people out of a job.
It's not likely to see a woman in woman in a black fur coat.
I can't decide whether a child or an adult dies.
It can't see at 400hz like your eyes can.
[Yes, we have persistence of about 16-24 hz, like you though you did, but we can see a new object enter the scene at about 400hz. Try it with an Aduino if you don't believe me. You can plainly see the difference between 60-120hz in monitors.]

Re:Who will be the first to die by it?

[wonkavader]

You only see well in your fovea. You may see movement in your peripheral vision, but you have no idea what it is. It sees well in 360 degrees.

Humans think slowly. If you have time enough to decide whether the baby or the adult dies then, in the same situation, the car should have plenty of time not to hit EITHER of them.

And it's taken WAY more driving tests than we have.

Re:Who will be the first to die by it?

[Jack Griffin]

You only see well in your fovea. You may see movement in your peripheral vision, but you have no idea what it is. It sees well in 360 degrees.

Except buses, apparently.

Re:Who will be the first to die by it?

[Anonymous Coward]

>It's not likely to see a woman in woman in a black fur coat.
Sorry, we can't really equip the cars with sonographic medical equipment and even for the med stuff you'd have to drop the fur coat.

Re:Who will be the first to die by it?

[Anonymous Coward]

It never had to take a driving test like you did.

Your point being..?

It will put thousands of people out of a job.

Who will be put out of a job?

It's not likely to see a woman in woman in a black fur coat.

It has radar, lasers and stereo cameras. It'll see better than you can.

I can't decide whether a child or an adult dies.

Good on you, mate.

It can't see at 400hz like your eyes can.
[Yes, we have persistence of about 16-24 hz, like you though you did, but we can see a new object enter the scene at about 400hz. Try it with an Aduino if you don't believe me. You can plainly see the difference between 60-120hz in monitors.]

Even if that meant anything, it's still going to be able to detect things better than you can, due to it having a lot more, and a lot more powerful sensors than you do. Again, radar, lasers, and cameras.
http://www.google.com/selfdrivingcar/how/

Re:Who will be the first to die by it?

[Anonymous Coward]

It never had to take a driving test like you did.

Yes - it is taking the tests right now. It hasn't passed yet - which is why they still have to ahve a person inside ready to take over.

There is no way robots will drive autonomously unless they can pass a standard driving test. Surely, some jurisdictions will demand exactly that - a standard driving test (not necessarily for each robot, but for each robot version.)

As for being backed by a large company - traffic law is simple. Super-rich guys driving badly still gets tickets & fines. And they get sued after accidents because they have ability to pay out a lot if they loose.

Re:Who will be the first to die by it?

[Anonymous Coward]

It never had to take a driving test like you did.

What do you think all this testing is right now? The difference is that the software doesn't forget items like a 16 or 17 year old human does.

Re:Who will be the first to die by it?

[Anonymous Coward]

It never had to take a driving test like you did.

You're right, its driving test was nothing like any of us had.
It is required to prove that it can drive safely and reliably over literally millions of miles, for multiple years, and have intimate and detailed knowledge of all the traffic laws in every State in which it wishes to operate.
My test was less than 3 miles of driving, and took less than 20 minutes including the time to Parallel park, and a short multiple choice 'test' which had less than 50 questions, focused only on the driving laws within my specific State. Yet my license allows me to drive in every State in the country, as well as the US and Canada.

It will come defended by one of he largest companies the world has ever seen.

A company which has stated that it is more than willing to automatically assume liability for its vehicles, and which supports enacting Regulations which enshrine that into law.

It will put thousands of people out of a job.

Including asshole drivers like the one in this story who attempted to illegally pass a car within the same lane, and OTR truck drivers who routinely falsify log entries so they can drive on far too little sleep, kept awake only by massive doses of stimulants, often including Meth.

It's not likely to see a woman in woman in a black fur coat.

And also not likely to slam into the car in front of it while it stares at a women in a string bikini.

I can't decide whether a child or an adult dies.

You should probably see a counselor about your urges, I'm not sure how that relates to a vehicle.

It can't see at 400hz like your eyes can.

I'm not sure how that's even relevant, but I will point out that your eyes don't see in the infrared spectrum at all, nor are they equipped to handle wavelengths such as are used in various types of radar/lidar, and that at best your eyes will degrade over time while sensors on a car can be replaced, refined, and updated at will.

Re:Who will be the first to die by it?

[delt0r]

It can see in all 360 compared to a very narrow specific view.
It can see with radar, in rain hail and snow.
It has reaction times measured in millionths of a second compared to mere tenths of a second or even longer.
It does not get angry.
It does not get impatient.
it is never tired.
it *never* speeds with the mistaken idea that it is better than all other drivers.
it is never stoned or drunk.
it is never texting.
it is never adjusting the stero.
it is never distracted.
It can model car dynamics and adjust for road conditions 1000s of times a second.


It is hands down better at driving that a evolved monkey.

No such thing as wide lane ...

[perpenso]

So based on numerous descriptions I have read, the Google car was in a very wide lane and moved to the right side of the lane to make a right turn. It saw some sandbags blocking the very right side of the lane, so it tried to move back to the middle of the lane.

I don't think there is any such thing as a wide lane where cars are allowed to go side by side. Whether it is marked as such by paint on the road or not the google car had moved into a right turn lane. It is common for parking spots and bike lanes and such to turn into a right turn lane near a corner. So marked with paint or not, the google car seems to have being trying to move from a right turn lane to a traffic lane. Perhaps the lack of paint contributed to the error, the software failed to recognize the right turn lane. An implicit lane even with the lack of paint on the road.

Re:No such thing as wide lane ...

[Joey Vegetables]

It has been common for my city to try to narrow 4 and 5 lane roads in my area into 2, for "traffic calming" purposes. One of the first things they typically do is to remove the lane markings except for the double yellow line. The road remains, de facto, two or more lanes in each direction, but it is no longer marked as such. Drivers will typically continue to form multiple lanes. This is of course much more dangerous, but unavoidable; trying to occupy both lanes is *also* illegal and dangerous. (What you're supposed to do is what I do; see below.) Eventually they will paint a "bike lane" too narrow to be safe, and separate it from the side of the road *just* enough to prevent parking or a second lane of vehicular traffic. This lane will also zig-zag, again encouraging collisions. Sometimes it will abruptly end without warning, and reappear without warning, seemingly just to make it as dangerous as possible. Still, cars, especially in snowy or icy conditions when the lane markings can't be seen anyway, will treat it as it always was and rightfully still should be: a multi-lane road. There is no safe option anymore. What I typically do during the "no marking" or "invisible marking" stage is to obey common law: stay as far to the right as possible unless passing, and maintain sufficient stopping distance in case a bike, pedestrian, parked car, or other obstruction should suddenly appear. The tendency is for cops to ticket the faster and/or more reckless drivers, and/or those who risk or cause accidents; that is as it should be in those conditions, and since I try to optimize for safety, I am rarely among those who get those particular tickets. (I get plenty on the open road, where safety requires staying with traffic and thus exceeding the artificially low posted limits.)

Re:No such thing as wide lane ...

[Anonymous Coward]

Ever hear of lane-splitting? Especially in CA.

Re:No such thing as wide lane ...

[Reziac]

Sure to improve with this new trend of no painted lines on the road! /sarcasm

Re:No such thing as wide lane ...

[perpenso]

Sure to improve with this new trend of no painted lines on the road! /sarcasm

Keep in mind that snow often hides paint. Humans and software must be able to deal with unmarked lanes in many parts of the country. Including parts of California.

Re:No such thing as wide lane ...

[Anonymous Coward]

I don't think there is any such thing as a wide lane where cars are allowed to go side by side.

Ever hear of lane-splitting? Especially in CA.

Yeah, for motorcycles, not for cars.

Re:No such thing as wide lane ...

[Reziac]

Yes, I'm very familiar with what happens when paint is hidden or doesn't exist: most people take their half out of the middle.

Re:No such thing as wide lane ...

[Anonymous Coward]

I don't think there is any such thing as a wide lane where cars are allowed to go side by side

Correct. The purpose of the wide lane is so that once the car in front of you has actually started to enter the turn, you can keep driving even before the vehicle has fully cleared the lane.

Whether it is marked as such by paint on the road or not the google car had moved into a right turn lane.

No. If there's no marked lane, it's not a lane, regardless of how much people want it to be.

An implicit lane

Such a thing does not exist.

Re:No such thing as wide lane ...

[Anonymous Coward]

R
T
F
A

One lane.
Multiple cars in lane.
Google car wants to turn.
Google car moves into far right side of lane to get around traffic in said SINGLE lane (To my knowledge this is legal, but I live in NY, ymmv)
Google car sees sandbags placed around a drain in the way.
Google car stops.
Light turns green.
Multiple cars move.
Google car waits.
Google car notices no cars next to it. Bus approaching at slow speed.
Google car moves to left side of THE SAME SINGLE LANE.
Bus continues on path.
Accident occurs.

ALL ACTIONS HAPPEN IN ONE WIDE LANE.

It's possible the google car is driving in the shoulder, but again, depending on local law this is perfectly legal, and or non-enforced as it aids in traffic flow (i.e. allowing right turns to occur and not block up single lanes of traffic)

Re:No such thing as wide lane ...

[Anonymous Coward]

R T F A

Done long ago.

One lane

Wrong. The right turn lane is a distinct and separate lane. You are wrong to think paint is a requirement.

Google car notices no cars next to it. Bus approaching at slow speed.

Given that the google car impact the *side* of the bus it did not notice the bus next to it.

It's possible the google car is driving in the shoulder ...

Wrong. At corners, parking, bicycle lanes, shoulders, etc become a right turn lane.

Drive over the sand bags?

[SuperKendall]

I don't think so, those things are pretty big generally. I don't know many drivers that would go over them even in an SUV.... but certainly not in a tiny car.

This all misses the point

[wonkavader]

All the lead-up to the real problem is just noise. Here's the real take-away:

The google car had a sudden obstacle come up. It was only moving two miles an hour, and yet it did not get out of the way fast enough. Its reflexes were too slow.

This is PRECISELY what autonomous vehicles should be able to do massively better than humans.

Probably Moore's law will solve this in a few years, but we need them to be much faster than humans now, because this is how we keep people running out from between cars from getting killed by autonomous cars going faster than 2 miles an hour. We can't let them on the roads in numbers and on their own without this. One death and it's no autonomous cars for an extra decade. So Google needs to figure out how to make its collision avoidance stuff faster.

It's a huge problem, it's their problem, and they have to fix it now.

Re:This all misses the point

[ColdSam]

Are you assuming that they can't do this better than the average human right now? My opinion is that the best cars available already can.

So the question is do we delay a "better than average driver" for political reasons while knowing that it will cost many lives and countless accidents just because of the politics.It would be craven and selfish of Google to make this call on its own, it is a decision that should be made by non-gutless politicians. If we can find any.

Re:This all misses the point

[Jack Griffin]

Are you assuming that they can't do this better than the average human right now?

They can't do better than me, and when I'm thinking about my next car purchase, that is all that counts.

Re:This all misses the point

[ColdSam]

Because you, like all drivers, are so much better than the average driver. I'm sure your ancestors said the same things about every technological advance, fighting progress all the way. They would be proud of you.

Re:This all misses the point

[Jack Griffin]

Because you, like all drivers, are so much better than the average driver.

I've never crashed into a bus, so I'm doing better than this stupid robot car. I will also bet you my house that I can get to my destination quicker than it most of the time.
So yeah let's not play around with vagaries, get back to me when those very real, measurable metrics improve.

Re:This all misses the point

[cwsumner]

The idea that computers are faster than humans, in this sort of real-world problem, is a fantasy. The human brain is massivly parallel processing and is at least an order of magnitude faster than computers in this kind of thing.

Don't be fooled by "Superman is invulnerable" type of supersticious thinking that is common in the public media. It ain't true! 8-}

Re:This all misses the point

[ColdSam]

I will bet my house that 5>3. Now, if we are done with the nonsense of making up new propositions, let's get back to the original statement.

Are you willing to bet your house that in a simulator you will be able to detect all relevant vehicles and obstacles better than the current generation of Google AV cars? If you are then let's set that up, otherwise you are just talking nonsense.

In the meantime, look up the legend of John Henry.

Re:This all misses the point

[ColdSam]

You are misinformed. Computers are much better at many real world problems such as this today, whether it is tracking incoming missiles, bouncing balls or, of course, tracking multiple vehicles and roadway obstacles. This is not to say that these systems are better in every circumstance, but the number of cases where a human will detect and avoid such an obstacle and the AV car cannot are getting fewer every day. I look forward to seeing results of Tesla's auto drive (a simple system compared to full AV like Google) vs. real world drivers.

I expect that in the near future, tech museums and other installations will allow you to pit your skills against the computer in a simulator. This is what it will take to convince most people, although some like Jack Griffin, will just never give up their feeling of superiority.

Re:This all misses the point

[cwsumner]

That's true where the environment is reletively low noise, like in gunnery and missile tracking. And computer systems are good at sequential multistage math, and at maintaining attention and focus for long periods.

But in complex environments with muliple type of high noise sources, such as a busy city street, it is quite different. The computer systems can get very slow, very soon. The problem with the predictions of what will be done in the future, is that many making predictions have no idea of the complexity of the "real world"! 8-)

Re:This all misses the point

[ColdSam]

As I said, you can manufacture scenarios and conditions where human drivers are better. However, if you have seen any of the videos of what autonomous cars "see" and how they are tracking and identifying dozens of relevant objects, far more than most humans can effectively process. Humans are successful because they are good at weeding out extraneous data MOST of the time, but the many millions of reported accidents per year show they make mistakes quite often.

Would you trust Tesla's autodrive to take you on a 10 mile highway run or a completely random driver? I think you know what my answer is.

AVs are not ready to take over all driving, but as to the question that I originally referred to, I think they are better than the average human at the straightforward task of identifying obstacles in real time and avoiding them and I think the data back me up on that.

Re:This all misses the point

[Jack Griffin]

Are you willing to bet your house that in a simulator you will be able to detect all relevant vehicles and obstacles better...

No because I live in real life, not a simulator. And in real life I will detect a sandbag and a bus 100% of the time, while it's obvious the google car doesn't.
Feel free to send me the deeds to your house whenever you are ready...

Re:This all misses the point

[cwsumner]

True, computers are better at maintaining attention. But much of what humans do is sub-consious automatic procedures, that they don't even know they are doing.

Besides, humans are not that bad as drivers. Keep in mind that the news media report only the disasters, with no sense of proportion about how bad the problems are. In short, the news people lie to get viewers. And always have. The Autodrive cars are going to have to go quite well to really be better. Even taking account of the humans driving while reading!

What worries me is that there will be a lot of autodrive cars out there, before they find out they have bad problems. But at least Google is running test trips for a while, maybe they will learn some things.

Re:Google's cars made illegal right turns all alon

[SydShamino]

It was previously not practical to write the code to enable the feature, because they were still working out bugs in the not-riding-up-the-curb-and-hitting-pedestrians code. When that code was tested, it was practical to add new features.

I just made that up, but with a law poorly written so as to include an "as close as practical" clause, it's good enough.

Checkbox for autonomous mode or conventional mode

[burleywinz]

I found it interesting that the California DMV has a checkbox for autonomous mode or conventional mode in their traffic accident form.

Won't happen when ALL vehicles are autonomous

[arobatino]

It should be kept in mind that if both vehicles were autonomous, they could have automatically negotiated what to do before either moved an inch, and this type of incident would never happen.

Re:Won't happen when ALL vehicles are autonomous

[Jack Griffin]

It should be kept in mind that if both vehicles were autonomous, they could have automatically negotiated what to do before either moved an inch, and this type of incident would never happen.

Sweet so we'll all be moving at 2mph. Good luck selling that dream...

Re:Won't happen when ALL vehicles are autonomous

[arobatino]

At 55 mph, it takes roughly 1/1000th of a second to travel an inch. With radio communication and the speed of computer hardware, that's a long time. There wouldn't be any perceptible lag, unlike with human drivers, who can take seconds to move after the light turns green.

Re:Won't happen when ALL vehicles are autonomous

[Jack Griffin]

At 55 mph, it takes roughly 1/1000th of a second to travel an inch. With radio communication and the speed of computer hardware, that's a long time.

Assuming AI image recognition works as well as a human, which it doesn't.

There wouldn't be any perceptible lag, unlike with human drivers, who can take seconds to move after the light turns green.

Unless there was a sandbag on the road, then it would drive into a bus...

Re:Won't happen when ALL vehicles are autonomous

[arobatino]

Assuming AI image recognition works as well as a human, which it doesn't.

The vehicles tell each other explicitly where they are. They don't need AI to avoid colliding with each other (as in this particular accident). They only need AI to deal with the rest of the environment (and the environment the cars spend most of their time on will be structured to be easy to deal with). Also, if they wanted, pedestrians could carry beacons saying basically "I'm a person, don't hit me" in case the AI isn't reliable enough, or if certain children have a habit of running into the road too fast for vehicles to avoid hitting them (just due to time required to change speed/direction, not processing time). In the latter case, the "network" could keep track of the behavior of those particular people, so as to give them a wide berth. That degree of safety is way beyond what's possible with human drivers.

Re:Won't happen when ALL vehicles are autonomous

[Jack Griffin]

The vehicles tell each other explicitly where they are.

So suddenly this discussion has turned from, can a robot car work effectively, to every car must be a robot car for it to work, and all humans and pets and wildlife and loose rubbish must carry tracking devices to ensure the robot car works properly. Seriously, that is your sales pitch?

Re:Won't happen when ALL vehicles are autonomous

[arobatino]

No, it's that with robot cars, those options exist to improve safety, if people want to use them, and with human cars they don't exist at all. Without any of the communication between cars and pedestrians/animals, safety and speed would still be improved, just not by as much. Given that cars are currently required to have lights/horns/license plates, it's likely that robot cars would be required to communicate as well.

Re:Won't happen when ALL vehicles are autonomous

[Jack Griffin]

No, it's that with robot cars, those options exist to improve safety, if people want to use them, and with human cars they don't exist at all.

So how is the road death rate reducing then? Magic fairies? The US could reduce it even further with mandatory seat belts and helmets, but if you can't even get that across the line, how will you force people into robot cars?

Without any of the communication between cars and pedestrians/animals, safety and speed would still be improved, just not by as much.

I'm yet to see any evidence of this whatsoever. Sure I can imagine a robot car could be safer in some areas, simply by being over cautious. But it can't be faster and legal.

Given that cars are currently required to have lights/horns/license plates, it's likely that robot cars would be required to communicate as well.

Just like how Google tracks everything I do? Yeah no thanks...

Trains

[Anonymous Coward]

They also need to note that trains *NEVER* yield.

And I'm thinking, angled approach it's going to be difficult to distinguish a bus from a train.

No Yield

[freudigst]

Bus drivers yield to no man. Why would they yield to a machine?

Re:Google's cars made illegal right turns all alon

[Jack Griffin]

I expected better from Google.

Why? Most things Google do are utter crap. One or two decent products among hundreds of duds...

Buses don't yield.

[Anonymous Coward]

Buses yielding???!?! HAHAHAHAHAHA.... ow my sides.. Buses from where I'm from never yield to anybody or anything.

Re: Don't let the perfect become the enemy of the

[Anonymous Coward]

Get back to us when these cars have even a 10% reduction in accident rates, can drive on roads they haven't seen before, and operate in snow or heavy rain. Until then, kindly keep imaginary statistics in the orifice you pulled that 98% from.

Yeah..

[Anonymous Coward]

This is why I don't drive in the city.. there's always someone going 2 or 15 or 35 - and a blind spot, and a delivery truck or some other such crap double parked at the end of my lane. It's a lot easier to just take the light rail and read Slashdot instead.

ABS

[ishmaelflood]

Sorry, ABS has scarcely any effect on stopping distances, particularly in the dry, for cars. It does do a reasonable job of cycling around the peak braking friction of the tire, but compared with a moderately skilled driver (which is the problem) the difference in stopping distance is small. ABS is designed to permit you to steer and brake simultaneously, and also as a by product it handles split mu.

In the dry, even if you lock all four wheels, your stopping distance is around 120% of the best that can be achieved, and even better, you'll stop in a straight line. That's why your first instinct, to stomp on the brakes, wasn't such a bad idea, even before ABS.

Opportunism

[ChoosyBeggar]

For me, it's not a stretch to imagine the bus driver simply saw the name "Google" on the car, and decided to collide on purpose. If he feels sure he can prove he's "in the right," why not go ahead and do it? If without monetary gain, he may be getting satisfaction out of causing trouble for "those rich, techie a-holes"...

buses in seattle

[Anonymous Coward]

In the day in downtown seattle, a million big buses lined the downtown curbs, filling up who blocks of curb solid. If one pulls out in front of you and you hit it, guess who got cited. So, a special case ordinance. We might end up with a lot of those with self-driving cars.

Who's legally at fault?

[Anonymous Coward]

I'm just curious. If it had been a serious wreck, and the autonomous car was at fault, whom is liable for the damages? Google? Or BMW, or Ford, when they bring autonomous cars out?

Is This Correct?

[LifesABeach]

A Bus is going 15mph, next to a car from Google, the Google car is going 2mph. Something does not add up.

lane sharing

[orgelspieler]

Why not just change the program so the Google car doesn't sneak past cars using the parking lane, and instead wait for a proper right turn at the intersection? Is there some expectation for people to pull to the parking are to make a right turn? That seems odd to me. But I'm not from an area that has a lot of on-street parking. Either it was one lane, and Google was wrong to make a second right turn lane. Or it was two lanes, and Google was wrong to veer into the other lane. I'd like to root for Google here, but I just can't figure out how it wouldn't be their fault. I'm glad they owned up and already fixed the software.