Google's Self-Driving Car Causes First Accident, As Programmers Try To Balance Human Simulacrum And Perfection
from the get-out-of-my-lane,-Dave dept
Google's self-driving cars have driven millions of miles with only a dozen or so accidents, all of them being the fault of human drivers rear-ending Google vehicles. In most of these cases, the drivers either weren't paying attention, or weren't prepared for a vehicle that was actually following traffic rules. But this week, an incident report by the California Department of Motor Vehicles (pdf) highlighted that a Google automated vehicle was at fault in an accident for what's believed to be the first time.According to the report, Google's vehicle was in the right-hand turn lane in a busy thoroughfare in Google's hometown of Mountain View, California, last month, when it was blocked by some sand bags. It attempted to move left to get around the sand bags, but slowly struck a city bus that the car's human observer assumed would slow down, but didn't. All in all, it's the kind of accident that any human being might take part in any day of the week. But given the press and the public's tendency toward occasional hysteria when self-driving technology proves fallible, Google's busy trying to get out ahead of the report.
Google historically compiles monthly reports for its self-driving car project; and while its February report addressing this specific accident hasn't been made public yet, Google's been releasing an early look to the media. In the report, Google notes that just like humans, trying to predict another driver's behavior isn't always successful on the road:
"Our test driver, who had been watching the bus in the mirror, also expected the bus to slow or stop. And we can imagine the bus driver assumed we were going to stay put. Unfortunately, all these assumptions led us to the same spot in the lane at the same time. This type of misunderstanding happens between human drivers on the road every day.Live and learn. Or compute and learn. Whatever. If automated vehicles were going to cause an accident, it's at least good that this appears to be an experience (don't give city bus drivers the benefit of the doubt) programmers will learn from. The problem historically is that like so many technologies, people are afraid of self-driving cars. As such, automated vehicles can't just be as good as human beings, they'll have to be better than human beings for people to become comfortable with the technology seeing widespread adoption.
This is a classic example of the negotiation that’s a normal part of driving – we’re all trying to predict each other’s movements. In this case, we clearly bear some responsibility, because if our car hadn’t moved there wouldn’t have been a collision. That said, our test driver believed the bus was going to slow or stop to allow us to merge into the traffic, and that there would be sufficient space to do that.
We’ve now reviewed this incident (and thousands of variations on it) in our simulator in detail and made refinements to our software. From now on, our cars will more deeply understand that buses (and other large vehicles) are less likely to yield to us than other types of vehicles, and we hope to handle situations like this more gracefully in the future."
By any measure self-driving cars have been notably safer than most people imagined, but obviously there's still work to do. Recent data from the General Motors-Carnegie Mellon Autonomous Driving Collaborative Research Lab suggests that self-driving cars have twice the accidents of human-driven vehicles -- but again largely because people aren't used to drivers that aren't willing to bend the rules a little bit. Striking an acceptable balance between having an automated driver be perfect -- and having an automated driver be more human like -- is going to be a work in progress for some time.
Thank you for reading this Techdirt post. With so many things competing for everyone’s attention these days, we really appreciate you giving us your time. We work hard every day to put quality content out there for our community.
Techdirt is one of the few remaining truly independent media outlets. We do not have a giant corporation behind us, and we rely heavily on our community to support us, in an age when advertisers are increasingly uninterested in sponsoring small, independent sites — especially a site like ours that is unwilling to pull punches in its reporting and analysis.
While other websites have resorted to paywalls, registration requirements, and increasingly annoying/intrusive advertising, we have always kept Techdirt open and available to anyone. But in order to continue doing so, we need your support. We offer a variety of ways for our readers to support us, from direct donations to special subscriptions and cool merchandise — and every little bit helps. Thank you.
–The Techdirt Team
Filed Under: accidents, autonomous vehicles, blame, fault, self driving cars
Companies: google
Reader Comments
Subscribe: RSS
View by: Time | Thread
[ link to this | view in chronology ]
Re:
[ link to this | view in chronology ]
Re: Re:
[ link to this | view in chronology ]
Re: Re: Re:
The car/driver changing lanes is supposed to ensure that a lane change can be safely performed, including making sure that there's enough room. The lane-changer cannot assume that the other driver has seen them; that driver could be be keeping an eye one some other concern that the lane-changer can't see, say in the other lane.
Not to mention, that in some jurisdictions (Québec, to name one), buses always have priority, even when they're pulling into your lane. Vous avez été avertis.
[ link to this | view in chronology ]
Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re:
Someone doesn't have to be stupid or a bad driver to think lane splitting is a bad idea. On the other hand he did call you an idiot first...
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re:
Consider three things:
1) Lane splitting puts three motorists (the biker and the two cars he's going between) in a position where they have a greatly-reduced margin of error.
2) To err is human.
3) When the inevitable does eventually happen, (see first two points,) you're the only one of the three who is not protected by a couple tons of metal armor.
If that realization does not put you off lane splitting forever, then yes, you are an idiot.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: Re:
I hate having a motorcycle stuck in a lane around me when there is stop and go traffic and cars everywhere. It seems like the potential for hitting them from behind if they are in the axis of your motion is greater than if they are parallel next to you. There are a lot of arguments to be made on both sides, it might be useful to look at some data on how most motorcycle accidents occur and what the laws are in the areas that they occur.
But like car drivers there are good and bad motorcycle drivers. I don't you can uniformly say that lane splitting is always bad just like you can't uniformly say that passing up another car is bad. In some situations it may be more dangerous than others and it's up to the driver to make those decisions. Some people are more careful than others when they do it.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: Re: Re:
Lane-splitting is inherently a surprising thing: it means that every motorcycle on a multilane road now becomes a random factor that other drivers have to pay extra attention to, just in case they start driving where there are no lanes.
It violates the #1 driver safety rule: never do anything surprising.
[ link to this | view in chronology ]
Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re: Re: Re:
An example of this would be two drivers arriving at a 4 way stop at the same time, the person to the left is required to yield to the person to their right. Also, hitting someone while driving through a "yield" sign would easily qualify.
I am sure there are a number of other situations, but in the case of the Google car, the Google car failed to yield to the bus, not the other way around.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re:
Yield signs are a bit of a sore spot for me at the moment, because in Pennsylvania I see them all over the place in places where they should not be: at the end of on-ramps.
If you've ever been to a driver's ed course, you'll remember that the purpose of an on-ramp is specifically to give you space to get up to speed and merge onto the highway safely. But around here, the civil engineers appear to have failed to understand that: instead of continuing for a reasonable distance (ie. at least half a mile), the lanes provided by most on-ramps vanish right after they meet up with the main highway, with a big YIELD sign there, which is dangerous (it's only safe to merge if you're going approximately the same speed as traffic in the lane you're merging into, and yield can potentially mean having to come to a complete stop with no more room to accelerate!) and defeats the entire purpose of having the on-ramp in the first place.
Mind you, I've got nothing against Yield signs used well. They have a legitimate purpose. I just don't see very many of them used right anymore.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re: Re:
http://www.leginfo.ca.gov/cgi-bin/displaycode?section=veh&group=21001-22000&file=21800-2 1809
[ link to this | view in chronology ]
Re: Re: Re: Re:
There are plenty of times you're required to yield, and failing to do so is a violation.
[ link to this | view in chronology ]
Re: Re: Re: Re:
Usually the laws address things like failure to yield to pedestrians or emergency vehicles. There are also some out there that specifically address roundabouts and traffic circles.
[ link to this | view in chronology ]
Re: Re: Re:
[ link to this | view in chronology ]
Re:
[ link to this | view in chronology ]
Where in the world are you from, where that looks like an everyday occurrence? I've lived all over the US and also outside it, and I don't believe I've ever seen sandbags in the road obstructing traffic, particularly in the middle of "a busy thoroughfare"!
[ link to this | view in chronology ]
Re:
[ link to this | view in chronology ]
Re:
I've seen sandbags in the middle of the road where I live more than once. They are used to weigh down temporary construction signs and sometimes do not get picked up with the sign.
[ link to this | view in chronology ]
[ link to this | view in chronology ]
Re:
[ link to this | view in chronology ]
Re: Re:
[ link to this | view in chronology ]
Re: Re: Re:
If something is enforced to make money, it has to be extremely clear where the lines are drawn and thought has to be given as to where the money goes since unclear enforcement and conflict of interests are primary recipies for corruption down the line.
[ link to this | view in chronology ]
Re: Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re:
[ link to this | view in chronology ]
Re: Sure, program the car to DELIBERATELY break the law. That will go over well.
Frankly, I can see this becoming an ideological issue, where one side stubbornly insists on strict enforcement of certain laws, refusing to acknowledge accumulating evidence of the wrongness of that course. Does that sound familiar?
[ link to this | view in chronology ]
The media sure does love to exaggerate.
[ link to this | view in chronology ]
Sorry, but no. I like Google, but I've gotta fault them on this. Any competent driver knows to never assume another vehicle will not do something stupid.
When I'm about to change lanes and there's any possibility of another vehicle entering the same space, I put my turn signal on and turn the wheel just a tiny bit, drifting over slowly so the other driver can get the message. Then I watch the other car, and if they don't clearly defer to me and make room for me to move in, I abort, pull back to the center of my lane, cancel the turn signal... and then pull in behind them and honk at them for being a jerk who doesn't know to defer to someone with a turn signal on.
But I never just go and assume they'll make room for me before they actually make room for me. That's just asking to get in a wreck, as we see here.
[ link to this | view in chronology ]
Re:
But the real important thing here is that they could analyze the crash (and thousands of variations on it) to prevent a similar accident from occurring. Unlike human drivers, which will cause the same accidents over and over, self-driving cars can learn not to repeat past mistakes.
[ link to this | view in chronology ]
Re: Re:
[ link to this | view in chronology ]
Re:
[ link to this | view in chronology ]
Re:
Defer? In my state it is the person who is changing lanes responsibility to make sure the lane is clear. There is nothing in our laws that says other motorists have to "defer" to your turn signal, adjust their speed or even make room for you.
[ link to this | view in chronology ]
Re: Re:
[ link to this | view in chronology ]
Re: Re: Re:
I see it the other way. The person changing lanes, expecting other drivers to "defer" to their turn signal is the one being discourteous. Why should I have to accommodate some idiot who doesn't know which lane they should be in?
[ link to this | view in chronology ]
Re: Re: Re: Re:
Anybody changing lanes is an idiot?
[ link to this | view in chronology ]
Re: Re: Re: Re: Re:
No. People who drive as if thier destination and thier time is infinitely more important than anyone else on the road with them is a pet peeve of mine. Using your turn signal as if it's a command signal that others should defer to, instead of a signal of your intentions is indictative of that sort of driver to me. Those are the idiots I was referring to.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re:
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: Re:
I agree and I do exactly that when it feels like I was asked. It's when I feel like I was commanded or just automatically expected to yield that I tend to be a bit of an *asshole. Courtesy is a two-way street.
* It might be that I really am an asshole, since I often have to resist the urge to park in handicapped spaces just to watch handicapped people make handicapped faces.
[ link to this | view in chronology ]
Re: Re: Re: Re: Re: Re: Re: Re:
That gave me a laugh....
https://www.youtube.com/watch?v=UrgpZ0fUixs
[ link to this | view in chronology ]
Google changes rules
[ link to this | view in chronology ]
Re: Google changes rules
Practice: That multi-ton hunk of metal heading your way is obeying a higher law: Newton's First. Failure to recognize it will result in your obeying Newton's Third.
[ link to this | view in chronology ]
Re: Re: Google changes rules
[ link to this | view in chronology ]
Re: Re: Re: Google changes rules
[ link to this | view in chronology ]
Re: Re: Google changes rules
1 - Bigger vehicles always have the right of way.
2 - More expensive cars always have the right of way.
3 - Piece of crap cars always have the right of way.
Why? Number 1 is obvious - they don't allow folks in because they don't have too; they'll win any kind of collision. Number 2 is because all expensive cars come with a certificate that says the owner can break any laws they feel like because they have money. And finally, number 3 is because the guy in the piece of crap just doesn't care - he paid jack for the car, and jack is what you'll get from him in any kind of accident.
[ link to this | view in chronology ]
Re: Re: Google changes rules
[ link to this | view in chronology ]
Re: Google changes rules
If you're a boater you should probably review the rules, because that ain't the rules.
http://www.boatus.org/guide/navigation_1.html
[ link to this | view in chronology ]
Re: Google changes rules
[ link to this | view in chronology ]
Re: Google changes rules
[ link to this | view in chronology ]
Re: Re: Google changes rules
Not that that doesn't happen, but don't be too quick to assume malice. The person could be completely oblivious to your presence. It's amazing how often drivers are unaware of the vehicles around them.
[ link to this | view in chronology ]
Re: Re: Re: Google changes rules
[ link to this | view in chronology ]
Needs of the many or needs of the few?
[ link to this | view in chronology ]
Re: Needs of the many or needs of the few?
[ link to this | view in chronology ]
Not to be trusted
Their behavior around the Tor debacle was duplicitous and disgraceful. Not that they did the research, but that they denied it. They should have done the research while keeping Tor informed of their findings in near real time. If that lost them the government money, so be it.
[ link to this | view in chronology ]
The unspoken bargain.
[ link to this | view in chronology ]
Re: The unspoken bargain.
No, I really can't. There's another exit following the one they failed to plan for. The right thing for them to do is not to put other drivers at risk (however "minor"), but to use the next exit instead.
[ link to this | view in chronology ]
patches
[ link to this | view in chronology ]
Re: patches
[ link to this | view in chronology ]
[ link to this | view in chronology ]
i'm pretty good at paper over the truth. if they pay well enough i might pitch in.
[ link to this | view in chronology ]
I actually have a lot more sympathy for the self-driving car in this case (bus should have yielded but did not)* than I do when it gets rear-ended while driving 25 when the speed limit is 45.
* - I formed my opinion on this primarily from the ars technica article, which described the lane positions somewhat differently.
[ link to this | view in chronology ]
Re:
According to the report, at one point the car "moved to the right-hand side of the lane to pass traffic in the same lane". Also, the bus was doing all of 15 MPH and was struck in the *side*. Perhaps that will inform your opinion as to what the car was doing vs what the bus was doing, and who was at fault.
[ link to this | view in chronology ]
Also if someone is on their cell phone other surrounding drivers may see that and compensate by being extra careful. If someone sees another car swerving a human may either anticipate that the car ahead is swerving around something, so they will then anticipate a potential obstacle ahead and react accordingly, or they will anticipate that the other driver is perhaps intoxicated, and they will attempt to compensate. How well do automated cars do this?
Also humans tend to avoid being in the blind spots of other vehicles for too long, especially large ones. They are pretty good at knowing if they are in a blind spot or not (if you can't see the other driver's face or its reflection). How well do automated cars do this?
And if a car in an adjacent lane stops it could be that someone is crossing the street so you should be more careful of pedestrians. People tend to know this. Do automated cars?
Things like accident avoidance are very important things to consider when driving and not just how well a car can follow a set of rules to avoid being legally at fault.
[ link to this | view in chronology ]
Re:
Hmmm. I must disagree. Besides, autonomous cars will probably be able to talk to each other in milliseconds and notice the "honks/expressions" much faster among themselves.
How well do automated cars do this?
As they learn to identify distracted drivers this will most likely improve. The factor that makes driving insecure and causes accidents is actually the human factor. We shouldn't be making predictions, we should just follow the rules and drive defensively and giving preference to those ahead of us.
Blind spots are not an issue to automated cars. Their blind spots will be as wide as their sensors precision. As for when dealing with human drivers the cars can probably evaluate such blind spots much better than humans. Maybe the cars haven't learned or haven't been taught this yet.
And if a car in an adjacent lane stops it could be that someone is crossing the street so you should be more careful of pedestrians. People tend to know this. Do automated cars?
Actually people are incredibly bad at it. I try to signal with my arm when I'm in the right lane to help but I've seen accidents almost happen because other drivers simply couldn't care less (or are distracted enough) that I actually think it's better to just keep flowing till there's a gap in traffic where drivers can see from afar that pedestrians are trying to cross.
Things like accident avoidance are very important things to consider when driving and not just how well a car can follow a set of rules to avoid being legally at fault.
That's what they are doing here. You see that both the car automated system AND the person inside thought the bus would slow down. A person that could have taken the controls if needed. Note that the vehicle moved slowly so it shouldn't have been a problem if the human inside decided to take control. It's not the AI fault. It's just the human factor.
[ link to this | view in chronology ]
Re:
[ link to this | view in chronology ]
Re: While driving people also communicate by hand gestures, facial expressions, and horn honking among other methods
[ link to this | view in chronology ]
[ link to this | view in chronology ]
Apparently the lanes are wide enough to allow 2 cars if the car turning right hugs the right of the lane, which the google car was programmed to do as a courtesy to other road users (not blocking traffic for people who aren't turning right).
[ link to this | view in chronology ]
The cars need to consider these assholish behaviors to avoid collisions. Maybe merge slowly and give full priority to human drivers or something like that. You can only program them to act strictly within the laws if there aren't humans driving in the vicinity. Obviously if there are only autonomous cars they can communicate among themselves and this would make things even easier.
[ link to this | view in chronology ]
lol
God DOES have a sense of humor.
[ link to this | view in chronology ]
Better
That's a fairly low bar in the US.
[ link to this | view in chronology ]
as in not moving.
[ link to this | view in chronology ]
Re:
This is not actually a real problem.
[ link to this | view in chronology ]
[ link to this | view in chronology ]
[ link to this | view in chronology ]