Apple Records First-Ever Accident In Self-Driving Car Program (appleinsider.com)

Apple's self-driving car program has reported its first-ever accident, according to a filing to the state's DMV. No injuries were reported. AppleInsider reports: A test car was rear-ended by a Nissan Leaf while merging onto an expressway, Bloomberg's Mark Gurman said on Twitter. The Apple vehicle suffered "moderate" damage. Details are still forthcoming, so it's unclear if the fault was with the Nissan driver, Apple's hardware and software, or some combination of the two. In an update, AppleInsider provided the following information: "The Apple vehicle, a Lexus SUV, was merging onto the Lawrence Expressway in California's Bay Area on Aug. 24, Gurman later wrote, citing a filing by Apple's Steve Kenner with the Department of Motor Vehicles. The Leaf was moving at just 15 miles per hour, but was also damaged."

Comments

Other driver obviously at fault

[Anonymous Coward]

If you rear end someone it means you were following too closely and could not manage your speed appropriately.

Re:Other driver obviously at fault

[SeaFox]

The Leaf was only driving 15 mph. Sounds like the driver was looking somewhere besides what was right right in front of him (like maybe over his shoulder at oncoming traffic from up the way) and didn't notice the Apple vehicle had slowed down/stopped while waiting for an opening to merge.

Re:Other driver obviously at fault

[Guybrush_T]

Apple’s car was traveling at less than 1 mile per hour, while the car that rear-ended it, a Nissan Leaf, was moving at about 15 miles per hour

Merging on Lawrence Expy at 1 mph ??? Either it was congested and the other car driver was distracted, or the Apple car was doing something really dangerous ... the other car braked and ended up crashing at 15 mph (down from ?).

Re:Other driver obviously at fault

[Anonymous Coward]

That is correct -- at least in general.

However it seems likely that we will see more accidents where a self driving car is rear ended by a non-self driving car (after factoring out the increased number of non-self driving cars that have some form of collision avoidance built into them). Self driving cars are just going to behave differently than human piloted ones - esp. with respect to detecting a hazard that actually doesn't exist and braking unexpectedly. This, in fact, is apparently somewhat why the Uber self-driving car that killed the jaywalking pedestrian had it's "sensitivity" to objects dialed down -- when it was turned up, it resulted in a jerky ride as the car reacted to stuff that wasn't a problem at all.

If someone jams their brakes on for no reason on a freeway traveling at the speed limit, they may actually be at fault in a civil suit for the massive pileup that may result. This is called brake checking and is unsafe practice.

I'm guessing, knowing these merge lanes on Lawrence, that the Lexus appeared to the driver behind to have identified "its slot" and appeared to be taking that slot. At that point, the following driver probably began to devote more attention to identifying a slot they were going to squeeze into instead of to the Lexus because they were under the impression that the Lexus would continue its merge in a "human expected" fashion but the Lexus got "nervous" and aborted the merge and stopped -- but the driver behind was looking over to his left, not straight ahead. The driver in the back is, of course, at fault.

There's an added twist on a good portion of Lawrence -- the RIGHT lane is a car pool lane and that's what you merge into. If you're not a carpool, you have a fairly limited opportunity to get over one lane before you're subject to an expensive moving violation. During moderate rush hour it can be hard to get over in time. This motivates the person doing the merge to consider additional factors which is an additional distraction.

I think the merge problem in moderately heavy rush hour traffic will, in general, be an Achilles' heel for self driving cars. There's actually few opportunities to "safely" (applying the law of physics, the requirement not to interfere with a car which has the right of way on the main road, and the need for self-driving cars to be risk adverse for PR and legal reasons) merge -- there simply are no slots available to "safely" (leaving required distance in front and behind the cars that the self driving car merges between) because cars are already traveling "too close" to each other and certainly not over twice the safe distance from each other for the speed they are traveling. This is likely to result in the first self driving car on each entrance as rush hour picks up simply stopping at the end of the merge lane (making it, of course, even harder to eventually merge) and sitting there until rush hour is over and traffic is light enough to allow it to come from a dead stop to a safe merge with no merge lane left for acceleration. Of course, there is a servo effect here -- once every entrance has ten or so self driving cars blocking it, traffic will lighten up and they will slowly trickle onto the freeway -- but the service streets will be a mess and a trip that would have taken one hour might take three hours.

Re:Other driver obviously at fault

[Anonymous Coward]

In *most*, not all, circumstances. A situation in which the lead car does not indicate that it is slowing/stopping (failed brake lights + no hand signal, for example) can be an issue in which the following car would not necessarily be wholly at fault.

Of course, there's also the trivial case of cutting someone off and slamming on the brakes, a not-uncommon form of insurance fraud in some locales.

Re:Other driver obviously at fault

[dgatwood]

I'm guessing it was one of those awful Sunnyvale intersections like Kifer eastbound or Central/Enochs eastbound, where you're lucky to be moving 1 MPH during certain times of day. :-)

Re:Other driver obviously at fault

[Richard_at_work]

Yeah, not always. In the UK there is a big problem with "crash for cash" which tries to capitalise on that the Amy of thinking, where fraudsters will overtake you, pull back in much too close and then slam their brakes on immediately, giving you no chance to avoid them. The fraudsters will then claim for all sorts of medical, psychological and other compensations from you.

The rise of dash cams is slowly bringing them under control.

Similarly here - we need to full details before we can be sure who is to blame, but a car merging into moving traffic might bear some responsibility, especially if the traffic they are merging into is moving significantly faster than they are...

We shall see what the outcome of the investigation is.

Re:Other driver obviously at fault

[Richard_at_work]

iOS autocorrect strikes again - I have no idea what's Amy's way of thinking is, but I do know the fraudsters are trying to capitalise on that *way* of thinking...

Re:Other driver obviously at fault

[DontBeAMoran]

Traffic isn't that bad until you have to drive in reverse.

Re: Other driver obviously at fault

[sodul]

Blaming my Apple twice, eh?

Re:Other driver obviously at fault

[drinkypoo]

Yeah, not always. In the UK there is a big problem with "crash for cash" which tries to capitalise on that the Amy of thinking, where fraudsters will overtake you, pull back in much too close and then slam their brakes on immediately,

They have that everywhere. The poorer the citizenry, the more there is, but it's a problem every place. If you don't have a dash cam (as you say later) then you're screwed. It's hard to prove that someone did that stuff on purpose otherwise.

giving you no chance to avoid them.

Unless they have a vehicle with a shorter stopping distance than you do, you had a chance to avoid them — start braking as soon as they come over. Which is what you do if you're driving defensively, to maintain a safe distance.

Still, having a camera is a great idea.

Re:Other driver obviously at fault

[Duds]

That is not universally true by any means.

Most notably a case a friend of mine had when they were slowing down for a queue and someone changed lanes and immediately stood on the brakes.

In the pre-dash cam era the case took 18 months to win but he won it.

Re:Other driver obviously at fault

[thegarbz]

If you rear end someone it means you were following too closely and could not manage your speed appropriately.

You're making a lot of assumptions. Sometimes you're just a completely distracted and useless git.

I got rear ended at a bridge opening once. I was at a red light. Car was standing behind me. I saw the bridge opening, the instructions are clear: turn your motor off, you're going to be there for 20min.

So I took my foot off the break, and put on the handbreak and switched my car off and got rear-ended.

The guy behind me tapping away on his phone just saw the red lights on my car turn off and assumed the traffic was moving and drove right into me.

Re:Other driver obviously at fault

[LinuxIsGarbage]

At least your iOS quotation marks worked.

Re: Other driver obviously at fault

[mrclevesque]

"Well it's not impossible that it's Apple car's fault...."

It's apple's fault because the safety driver did not take over when the AI was obviously not able to merge with the traffic in a competent fashion. This kind of problem happens with all companies, also during left turns, in both instances the complexity of what humans do when negotiating these situations is way beyond the scope of what these AI systems can now accomplish.

Re:Other driver obviously at fault

[gnasher719]

One website had a few videos. First one of a woman jumping from behind a tree onto the street, right in front of a car, but estimating the jump wrong and getting badly hurt. One was a car overtaking a truck, braking hard but not managing to cause a crash, going on the fast line and falling back, overtaking again and repeating the manouvre. With everything caught on the truck's dash cam.

Re: Other driver obviously at fault

[mrclevesque]

"So it's Apple's fault that the security driver didn't ram the Apple car into the other cars that wouldn't let it merge. Yeah, you make sense."

I don't know the specifics of what happened so yeah.

But at the level 'autonomous' cars are today I think it's fair to always at least partially blame corporations when they permit their cars to enter situations that are beyond their car's AI's capabilities.

Re: Other driver obviously at fault

[mrclevesque]

I think the idea is more than if someone going only 15 mph rear-ended them, it was likely a situation beyond anyone's capabilities except for the person that rear-ended them. Would you also blame the self-driving car's corporation if it got rear-ended while stopped at a red light? Assuming a self-driving car must be partially to blame if struck by someone driving incorrectly is faulty logic. All that said, I have "rear-ended" a car before when I was stopped at a red light, and the car in front of me decided it wanted to back up to go another way without checking if another car was behind it. Unlikely to be what happened here, though.

"it was likely a situation beyond anyone's capabilities except for the person that rear-ended them. Would you also blame the self-driving car's corporation"

These companies have special permissions to be on the road, they have rules to follow above and beyond what your average driver has to. They have contractual obligations to not interfere in any way with the safety and operation of the public transport system, which includes the obligation to not test any system that would put anyone at a higher risk of injury.

That said, these companies, considering the state of the technology, should not under any circumstances let their cars merge or make left turns unless the safety driver has taken over the operation of the car, if not they should at minimum be fined, and if involved in an accident they should also as an extra deterrent be held at least partially responsible for damages, which would not preclude any other vehicle involved in the accident from being fined if they bore some or even all of the blame.

I blame iTunes

[jfdavis668]

It was probably updating when the crash occurred.

Sounds kind of hilarious

[SuperKendall]

The thought of a Leaf damaging anything, at any speed (much less 15 MPH), kind of makes me laugh.

They must quantify "moderate" damage by cost, it was probably over $10 to re-paint the bumper on the Apple Lexus SUV.

That said even though rear-endings are normally the fault of the follower, I have to wonder if the Apple self driving car did not do something super un-expected...

Sometimes..

[fluffernutter]

Sometimes human drivers are just responsible.

Re:Sometimes..

[Kjella]

Sometimes human drivers are just responsible.

Well you'd think so, considering they're driving. I suppose there can be a few mechanical faults where they're not, but otherwise duuuuh.

Re:Sometimes..

[fluffernutter]

No that's not true. If the robot car was doing something that a human never would have, I blame the robot car; because technology companies should be striving to make them act human so that other humans don't get confused. But unless the robot car was doing something unusual in this case, it is truly the human's fault.

Re:Sometimes..

[drinkypoo]

No that's not true. If the robot car was doing something that a human never would have, I blame the robot car; because technology companies should be striving to make them act human so that other humans don't get confused.

Name something that a robot car might do in traffic that a human would never do.

Re:Sometimes..

[thegarbz]

Sometimes human drivers are just responsible.

That's rarely the case. I tend to find human drivers mostly irresponsible.

Re: Sometimes..

[fluffernutter]

That, or get confused about what lane it is in and merge in front of a bus going faster than it, as a Waymo car did last year.

Re: Sometimes..

[drinkypoo]

Name something that a robot car might do in traffic that a human would never do.

Slam on the brakes erratically for no reason.

It wasn't for no reason, computers don't do things for no reason. It may have been a bad reason, but guess what? Human drivers do that shit all the time. So again, name something that a robot car might do in traffic that a human would never do. You haven't achieved it so far.

Which is why they turned down the sensors on the car they used to murder that pedestrian woman.

They didn't "turn down the sensors", they turned down the reactiveness. The vehicle's sensors saw the woman who stepped out in front of the vehicle in dark clothing at night in between crosswalks. Which, by the way, made it not murder. It made it the pedestrian's fault, which is what it was.

Re: Sometimes..

[fluffernutter]

If a hundred humans wouldn't have slammed on the breaks in the same situation, then it is safe to say that the reason the robot car is doing it for is not valid.

Re: Sometimes..

[fluffernutter]

Again.. it depends on what 100 humans would do in the same situation. You can't say humans would have done the same thing without testing it.

Re: Sometimes..

[fluffernutter]

Oh brother, yes people are idiots. It's a wonder we get rockets into the air, or perform successful surgeries.

Re: Sometimes..

[Quirkz]

Duh. Slam on the brakes erratically for no reason.

I see human-driven cars do this regularly. Frequently the actual reason is they're turning, but they didn't bother to put on their signal, or not until long after they'd hit the brakes.

Expressway

[dkman]

Nothing says merging onto the expressway like "going 15 mph"

Re:Expressway

[pipedwho]

It's worse than that. The Apple car was probably merging at 1mph while the guy behind sped up to 15mph to fill a gap in the next lane over, but mistimed it. That part of Lawrence Expressway is like a huge parking lot during 'peak hour(s)' which lasts for a good half of the day, morning and evening.

Re:Prediction: Applefans will defend the car..

[Anubis IV]

Including the next time Apple kills somebody with one.

While we’re on the topic of loaded phrasing with no basis in reality, when exactly did you decide to stop beating your wife?

Re:Come on, this is clickbait.

[mlyle]

> It's always the person in back's fault in a rear end collision. ALWAYS.

Not necessarily in a merge. I agree it's probably the human driver's fault, but if you dive in front of someone it is not necessarily physically possible for them to stop.