Slashdot Mirror
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.
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.
>> 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.
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?
Where are we going and why are we in a handbasket?
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.
All my liberal friends think I'm a conservative, all my conservative friends think I'm a liberal.
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.
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.
Knowledge Brings Fear
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.
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.
- (c) 2018 Hank Zimmerman
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.
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.
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.
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.
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.
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.
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.
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.
That that is is that that that that is not is not.