Automatic emergency braking should become mandatory, feds say

The rule would save 360 lives and prevent 24,000 crashes a year, NHTSA says.

See full article...

 

[Fatesrider]

But it's not entirely perfect. NHTSA wants standard automatic emergency braking systems to include pedestrian detection, but a 2019 study by the Automobile Association of America found that higher speeds or low light conditions seriously degraded performance. An overactive emergency braking system is undesirable, too—both Tesla and Honda have been investigated by NHTSA as a result of "phantom braking" complaints, where oversensitive or badly designed systems throw false positives, sometimes causing a crash in the process as the cars brake inappropriately.

I would hope in the course of making such a feature mandatory that they'd include minimal standards for efficacy, accuracy and reliability, and set them all fairly high, in the process.

 

[DistinctivelyCanuck]

So, is there a planned coordination between the article on software defined cars, and a feature which absolutely depends on well written and well tested software? LOL

Emergency braking can be exceptionally dangerous when it triggers for absolutely no reason whatsoever and results in being rear ended by the vehicle behind you which had zero expectation of a sudden braking event.

(asks the guy who more than once in these forums has complained vigorously about the poor quality of the Honda CR-V's driving assist suite)

 

[mihoda]

As someone who worked on real-time classifications that were under constant evaluation...

• Who is responsible for false positives?
• Who is responsible for false negatives?

 

[Dr Gitlin]

So, is there a planned coordination between the article on software defined cars, and a feature which absolutely depends on well written and well tested software? LOL

Emergency braking can be exceptionally dangerous when it triggers for absolutely no reason whatsoever and results in being rear ended by the vehicle behind you which had zero expectation of a sudden braking event.

(asks the guy who more than once in these forums has complained vigorously about the poor quality of the Honda CR-V's driving assist suite)

The only coordination is Eric messaging me while I was at lunch to let me know the AEB NPRM just dropped.

Of course, you’re free to imagine nefarious conspiracies if you like.

 

[jfconde]

As someone who worked on real-time classifications that were under constant evaluation...

• Who is responsible for false positives?
• Who is responsible for false negatives?

Obviously the driver is to be blamed for anything the systems do wrong, or fail to do right. Isn't that the pattern for all current "driver assists"?

 

[cerberusTI]

So, is there a planned coordination between the article on software defined cars, and a feature which absolutely depends on well written and well tested software? LOL

Emergency braking can be exceptionally dangerous when it triggers for absolutely no reason whatsoever and results in being rear ended by the vehicle behind you which had zero expectation of a sudden braking event.

(asks the guy who more than once in these forums has complained vigorously about the poor quality of the Honda CR-V's driving assist suite)

What kind of fool drives without expecting a sudden breaking event by the car ahead?

That is partially a technology failure, but the driver behind who failed to keep enough distance to manage is primarily at fault. That is the easily controllable part, and you need the ability to stop for a variety of reasons beyond a false positive emergency braking.

 

[Edgar Allan Esquire]

So, is there a planned coordination between the article on software defined cars, and a feature which absolutely depends on well written and well tested software? LOL

Emergency braking can be exceptionally dangerous when it triggers for absolutely no reason whatsoever and results in being rear ended by the vehicle behind you which had zero expectation of a sudden braking event.

(asks the guy who more than once in these forums has complained vigorously about the poor quality of the Honda CR-V's driving assist suite)

Theoretically, if enough cars have automatic brakes, a phantom event/false positive would be mitigated by the following cars also braking... assuming people actually maintain an appropriate follow distance instead of that 0.5-1 second window.

Although, right now my big concern is the sharp rise in cars I'm seeing drifting into oncoming traffic that almost always seems tied to fiddling with the center console infotainment tablet.

 

[atlcomputech]

My Ford Bronco all the sudden decided that there was an object in front of me on the highway and decided to hard brake at 70mph. It was scary and dangerous as hell.

 

[Omega_Prime]

Mandating an override of driver control would be highly problematic. I expect significant pass the buck on the liability side.

 

[einnocent]

If there is an overall lesson to be learned from the human factors literature, perhaps it is this: automation has a kind of totalizing logic to it. At each stage, remaining pockets of human judgment and discretion appear as bugs that need to be solved. Put more neutrally, human intelligence and machine intelligence have a hard time sharing control. This becomes evident in the problems posed by partially autonomous cars, and is evident also in the problems posed when fully autonomous cars have to share the road with human drivers.

Driverless cars are programmed to follow traffic rules to the letter and err on the side of caution, making them an awkward fit with other cars piloted by humans. The New York Times reports that one Google car “couldn’t get through a four-way stop because its sensors kept waiting for other (human) drivers to stop completely and let it go. The human drivers kept inching forward, looking for the advantage—paralyzing Google’s robot.” Of course, what human drivers do is make eye contact in such a situation, or read other cues of social interaction, allowing them to negotiate ambiguous cases of right-of-way and work things out on the fly. Some drivers are more assertive, others more defensive. It is not a stretch to say that there is a kind of body language of driving. This improvisation works just fine, for the most part.

But social intelligence is hard to reproduce with machine-executable logic. Therefore, it is concluded, human beings must become more like machines, in order to make the road more hospitable to robots. According to the same Times article, “Dmitri Dolgov, head of software for Google’s Self-Driving Car Project, said that one thing he had learned from the project was that human drivers needed to be ‘less idiotic.’” Such an inference comes easily when you conceive reason as a computer scientist does; as asocial and fundamentally rule-like. From such a perspective, human beings do indeed look like inferior versions of computers.

The contest for control between humans and computers often looks like no contest at all, as a political reality. Viewed through the lens of press releases and credulous journalism, the logic of automation is joined, in the public mind, to the moral logic of safety, which similarly admits no limit to its expansion. These two are symbiotic in the sense that safetyism provides legitimation to the business logic of ever more automation. Together they are unimpeachable in the minds of all right-thinking people, and to question Team Progress is to invite being labeled pro-death.

If we confine ourselves to factual questions about the likely safety gains to be had from self-driving cars, it seems to me the most prudent position would be to join the agnosticism of the human factors researchers. As Casner et al. say, “In the coming decades, we will all participate in driving research as an enormous uncontrolled experiment takes place on our streets and highways.” This they state as a matter of fact, as though we had no choice in the matter. Automaticity becomes a political mood, no less than an engineering project.

Crawford, Matthew B.. Why We Drive (pp. 104-106). HarperCollins.

 

[riverstyx]

You know what would really save lives? If NHTSA did their damn jobs and regulated giant 10 foot tall trucks and SUVs people use as mall crawlers maybe we wouldn’t be one of the worst in the developed world in terms of motor vehicle fatality rates, worse than our neighbors to the north. They also have a habit of constantly blaming pedestrians for deaths when every other developed nation is redesigning roads, highways and other critical infrastructure to reduce pedestrian fatalities. Yet all they’ll do is things like this and drum it up as some huge victory.

 

[schnackenpfefferhausen]

The AEB on my Volvo works quite well (the front collision warning is prone to false positives.. but it doesn’t engage the brakes).

It has activated one time. And I think it prevented a collision in that instance. It was a very chilling and sobering drive home after that (as well it should be).

 

[ip_what]

Mandating an override of driver control would be highly problematic. I expect significant pass the buck on the liability side.

We’ve had this since 2015. Everything from ABS to stability control and even lane keeping assist is capable of overriding driver control. It’s fine and we know how to deal with it. None of these technologies are perfect, but they do save lives. That’s more important than a direct link from driver to road for the vast majority of use cases.

 

[S4WRXTTCS]

What kind of fool drives without expecting a sudden breaking event by the car ahead?

That is partially a technology failure, but the driver behind who failed to keep enough distance to manage is primarily at fault. That is the easily controllable part, and you need the ability to stop for a variety of reasons beyond a false positive emergency breaking.

An automatic emergency braking event is a serious reduction in speed designed primarily to reduce the severity of an accident. Sure some systems are able to avoid the accident all together, but I think most on the road are severity reduction ones.

These are such severe reduction in speeds that no one behind a vehicle that does this is expecting that serious of speed change.

People often confuse phantom braking by Adaptive Cruise control systems for AEB braking, and its simply not the same thing. If someone runs into a phantom braking adaptive cruise control system then they are an idiot, but not a false automatic emergency braking event.

When I had a Tesla I experienced numerous phantom braking events by adaptive cruise control, but zero AEB events (either false or true).

The only time I've ever experienced an AEB braking event was in my Jeep Wrangler as I was entering into an underground parking garage. It was false triggered due to the wall before the drop. It was a very intense 20mph to zero.

 

[tRexx]

Make it mandatory for 360 lives a year? For a little perspective, we just lost over a million from Covid 19.

 

[ranthog]

An automatic emergency braking event is a serious reduction in speed designed primarily to reduce the severity of an accident. Sure some systems are able to avoid the accident all together, but I think most on the road are severity reduction ones.

These are such severe reduction in speeds that no one behind a vehicle that does this is expecting that serious of speed change.

People often confuse phantom braking by Adaptive Cruise control systems for AEB braking, and its simply not the same thing. If someone runs into a phantom braking adaptive cruise control system then they are an idiot, but not a false automatic emergency braking event.

When I had a Tesla I experienced numerous phantom braking events by adaptive cruise control, but zero AEB events (either false or true).

The only time I've ever experienced an AEB braking event was in my Jeep Wrangler as I was entering into an underground parking garage. It was false triggered due to the wall before the drop. It was a very intense 20mph to zero.

From my understanding most all systems will attempt to avoid an accident, but that they're not capable of doing so at all speeds, scenarios, and conditions. Which makes sense especially at higher speeds.

 

[total.wimp]

What kind of fool drives without expecting a sudden breaking event by the car ahead?

That is partially a technology failure, but the driver behind who failed to keep enough distance to manage is primarily at fault. That is the easily controllable part, and you need the ability to stop for a variety of reasons beyond a false positive emergency breaking.

You ever been on a busy freeway during rush hour? What people should do (give adequate room to stop) and what they actually do (drive way too close and pray someone doesn't suddenly stop in front of them) are two entirely different things. The problem is widespread. It would be great if our cars didn't exacerbate the the problem with sudden braking.

 

[theramenman]

Great. Let's make new cars cost even MORE thanks to tech that's not ready for primetime yet. After renting a Subaru with EyeSight that was overly ambitious, I think I'll pass and depend on myself to pay attention and hope the best for drivers around me. That Subaru slammed on its brakes when traffic was at a safe distance ahead but coming to a stop. It panicked like somebody who it was their first time behind the wheel. I know some people say it's a good system, so maybe mine was being glitchy as hell, but that makes it even spookier that these things CAN and will act up from time to time. I predict mandatory emergency braking could lead to MORE crashes, not less.

Most American drivers should not be on the road but our country refuses to follow international best practices so borderline blind grandma or the guy who failed his test 30 times and finally got his license both get to be on the same road as you. Either those morons can take out a pedestrian at a street crossing or an entire family in an SUV or you can add (however trigger happy) driver assistance features. Maybe you’re a great driver and don’t need this stuff but most aren’t. There are still a few basic, cheap, manual cars that no one (and I’d wager that includes you) buys. The most popular car in this country is a $40k truck. This isn’t gonna be the one feature that breaks the camels back in terms of pricing.

 

[Deleted member 853890]

Make it mandatory for 360 lives a year? For a little perspective, we just lost over a million from Covid 19.

Yes, this is TOTALLY analogous to a once in a lifetime pandemic! Let's get rid of seatbelts too, because those "only" save 15,000 lives a year!

 

[rhavenn]

It may save certain lives, but there is a good chance it will harm other lives. The sensors, capability, and , ultimately, the trust just aren't there. My 2020 Toyota will start to emergency break in a sharpish curve if a car is in the right spot is either moving or parked. I know I'm not going to hit it, but it starts screaming at me and once did start to emergency break but then stopped as the car passed by in the other lane like normal.

I live in the sticks, so I can only imagine what it would do in an actual city with traffic. The only thing I trust it do okay is the cruise control '"pace" mode where it'll pace the car in front and keep a certain distance, but that's on a divided and mostly straight as an arrow interstate. I would never use that in rush hour, real traffic or anything else that wasn't a divided highway.

By "trust" I mean both in the tech itself, the programmers and engineers building it and, ultimately, the corporations putting profits before anything else. aka: see the Tesla leaked docs and the shit they've been pushing under the rug. If these systems aren't open-source to at least the likes of NHTSA for review..then they can fuck right off. I don't care about data points and X million miles driven without issue. It needs independent code review because they will do their utmost to blame anything but their systems. Sometimes they're right that it was the driver at fault, but often times they're not.

 

[willdude]

What kind of fool drives without expecting a sudden breaking event by the car ahead?

That is partially a technology failure, but the driver behind who failed to keep enough distance to manage is primarily at fault. That is the easily controllable part, and you need the ability to stop for a variety of reasons beyond a false positive emergency breaking.

I dunno, I guess when I’m driving on like a highway where I can clearly see that the road beyond the car in front of me is clear, I’m not expecting them to suddenly slam on the brakes. Sure, they might if like a deer darts out into the road, in which case slamming on my own brake wouldn’t stop me in time. But if they brake for zero reason beyond an AEB false positive, I’d be pissed.

 

[S4WRXTTCS]

We’ve had this since 2015. Everything from ABS to stability control and even lane keeping assist is capable of overriding driver control. It’s fine and we know how to deal with it. None of these technologies are perfect, but they do save lives. That’s more important than a direct link from driver to road for the vast majority of use cases.

We have had this, but I wonder what the success rate of overriding an AEB event is?

It's such an intense reductions in speed that I don't think its physically possible to override. About all they can do is recover as quickly as possible from it.

False AEB events are pretty rare where if they do happen it leads to recalls.

Probably the greatest weakness in them isn't false positives, but negatives positives during the times where people need them the most. Like on rainy nights with pedestrians crossing the road without any reflective gear.

 

[ranthog]

Make it mandatory for 360 lives a year? For a little perspective, we just lost over a million from Covid 19.

Just because Republicans decided to wage a war on public health instead of being even mildly inconvenienced doesn't invalidate the need to do this. I agree it is a travesty that we didn't do more bout COVID-19 with how rich we are.

I support both, because doing both makes sense. The criminal lack of action (or incompentant action) on one does not justify ignoring the other issues that need to be taken care of.

 

[autostop]

I can't remember any time my car's radar+camera AEB has ever called a false-positive in three years' worth of driving. When it triggers, it's for good reason and I'm almost always braking along with it.

The purely camera-based lane-departure system, on the other hand, is somewhat prone to false positives sometimes when lanes diverge or markings are, requiring a steady hand on the wheel.

 

[reub2000]

You know what would really save lives? If NHTSA did their damn jobs and regulated giant 10 foot tall trucks and SUVs people use as mall crawlers maybe we wouldn’t be one of the worst in the developed world in terms of motor vehicle fatality rates, worse than our neighbors to the north. They also have a habit of constantly blaming pedestrians for deaths when every other developed nation is redesigning roads, highways and other critical infrastructure to reduce pedestrian fatalities. Yet all they’ll do is things like this and drum it up as some huge victory.

I think NHTSA should at the very least copy the European version of NCAP, which includes crash tests for pedestrian safety. Discourage vehicle designs with high hoods that endanger pedestrians needlessly.

 

[cerberusTI]

You ever been on a busy freeway during rush hour? What people should do (give adequate room to stop) and what they actually do (drive way too close and pray someone doesn't suddenly stop in front of them) are two entirely different things. The problem is widespread. It would be great if our cars didn't exacerbate the the problem with sudden braking.

There should be an effort to minimize it, yes. It is not great having your car stop on a freeway.

Still though, what if they blow a tire, or any other issue where the machine fails? Or if the driver has a medical issue? It is best to be able to keep enough distance to stop before you would hit the car ahead, even worst case.

In heavy traffic, that means paying extra attention so that you can cut it closer, and leaving as much of a gap as you dare before someone will slip through casually for reasons other than needing to get over. If you cannot stop, you are cutting it too close. The more distance you need to stop, the more careful and maybe more accommodating you should be (or travel outside rush hour).

 

[ranthog]

I think NHTSA should at the very least copy the European version of NCAP, which includes crash tests for pedestrian safety.

I support this as well. That is going to be a really big problem for the current design for a lot of vehicles, so its probably going to take years to implement after the rule is finalized. So sooner the better on this.

 

[watermeloncup]

Make it mandatory for 360 lives a year? For a little perspective, we just lost over a million from Covid 19.

Traffic fatalities, especially pedestrian fatalities have been climbing at an alarming rate in the US so I think it's about damn time they do something.

 

[AnimalMotherMF]

This is a horrible idea. The brake warning in my car goes off at the wrong times. If it was automatic the car would be stopping when I expect it to accelerate. There are too many times the stupid warning goes off when it should not. Currently it is an inconvenience and if it was automatic it would cause accidents.

 

[Deleted member 853890]

It may save certain lives, but there is a good chance it will harm other lives. The sensors, capability, and , ultimately, the trust just aren't there. My 2020 Toyota will start to emergency break in a sharpish curve if a car is in the right spot is either moving or parked. I know I'm not going to hit it, but it starts screaming at me and once did start to emergency break but then stopped as the car passed by in the other lane like normal.

I live in the sticks, so I can only imagine what it would do in an actual city with traffic. The only thing I trust it do okay is the cruise control '"pace" mode where it'll pace the car in front and keep a certain distance, but that's on a divided and mostly straight as an arrow interstate. I would never use that in rush hour, real traffic or anything else that wasn't a divided highway.

By "trust" I mean both in the tech itself, the programmers and engineers building it and, ultimately, the corporations putting profits before anything else. aka: see the Tesla leaked docs and the shit they've been pushing under the rug. If these systems aren't open-source to at least the likes of NHTSA for review..then they can fuck right off. I don't care about data points and X million miles driven without issue. It needs independent code review because they will do their utmost to blame anything but their systems. Sometimes they're right that it was the driver at fault, but often times they're not.

The rule wouldn't go into effect for four years after it's adopted, meaning (since 2023 models are already out and there are mandatory periods for comments and reviews before adoption) it'd be in 2028/9 model years if the wheels turn quickly and it's adopted soon. So, a decade of improvement past the version in your car.

 

[watermeloncup]

I think NHTSA should at the very least copy the European version of NCAP, which includes crash tests for pedestrian safety. Discourage vehicle designs with high hoods that endanger pedestrians needlessly.

The full sized trucks and SUVs are especially asinine. They have unnecessarily high hoods containing mostly empty space solely to make the drivers feel like a big man.

 

[barich]

You know what would really save lives? If NHTSA did their damn jobs and regulated giant 10 foot tall trucks and SUVs people use as mall crawlers maybe we wouldn’t be one of the worst in the developed world in terms of motor vehicle fatality rates, worse than our neighbors to the north. They also have a habit of constantly blaming pedestrians for deaths when every other developed nation is redesigning roads, highways and other critical infrastructure to reduce pedestrian fatalities. Yet all they’ll do is things like this and drum it up as some huge victory.

Look, sometimes people have extremely good reasons for having such high vehicles. Such as this example photo that I just snapped today:

 

[cerberusTI]

I dunno, I guess when I’m driving on like a highway where I can clearly see that the road beyond the car in front of me is clear, I’m not expecting them to suddenly slam on the brakes. Sure, they might if like a deer darts out into the road, in which case slamming on my own brake wouldn’t stop me in time. But if they brake for zero reason beyond an AEB false positive, I’d be pissed.

You could just give it some distance and not be in that position at all, why risk an accident if something goes a bit wrong ahead of you?

On a clear road, there is really no reason to be so close you would hit the car ahead if they hit a deer (or anything else went wrong).

 

[motytrah]

I think NHTSA should at the very least copy the European version of NCAP, which includes crash tests for pedestrian safety. Discourage vehicle designs with high hoods that endanger pedestrians needlessly.

They don't even require one peddle driving engage the brake lights. Which is why Hyundai/Kia ended up the way it did in the US.

 

[swiftdraw]

My Ford Bronco all the sudden decided that there was an object in front of me on the highway and decided to hard brake at 70mph. It was scary and dangerous as hell.

Same with my Sonata when it detected a empty garbage bag blowing across the highway and the guy riding my ass damn near rear ended me. I turned off the auto braking after that incident.