Hacker News Reader

Comment by JshWright | Hacker News Reader

Assuming the occupant didn't see it coming, the car will have made all its decisions and initiated all active safety measures before a human being inside the car would even notice a collision is taking place.

The car is effectively 'done' progressing the event before the human even starts.

View on HN

It actually seems a lot slower than I would have expected, to be honest. The seatbelt tensioners take 150ms to activate, during which time I would have travelled 2m at 30mph (1). That effectively makes them useless, no?

1) https://www.wolframalpha.com/input/?i=distance+traveled+in+1...

No, the pretensioner just has to beat the deceleration of the car. The goal is to retain (or even pull in) the passenger in the seat to reduce the impact on the airbags.

Typically the pretensioner will pull in on the order of 10-12cm of webbing, this is done via a pyrotechnic charge. To be fully effective, the pretensioner would then need to deploy about 110-120 milliseconds before the airbags (which will deploy in 40 ms or so.)

So then it falls on the car's sensors and computer to detect with sufficient advance notice the impending crash. With modern radars (like the ones in newer mercedes cars...) they should be able to gain most of that time at least for frontal crashes.

Depends on what kind of crash it is. My dumb guess is that most crashes can be predicted by sensors in advance of the moment of impact, perhaps by significant multiples of 150ms.

For example consider a crash where your car slips on ice and drives right into a tree. Before you hit the tree there's gonna be all kinds of crazy steering and braking attempts that sensors could detect, and the accelerometers are gonna show weird lateral ice-slipping motions.

> most crashes can be predicted by sensors in advance of the moment of impact,

Provided you dont have any false alarms. For instance, someone making an unsafe pass on a 2-lane road, lets say a fast motorcycle. radar sensor sees incoming object with relative velocity of 150mph, see's driver slam brakes, yet motorcycle dives back into his lane at the last second. In this case, deploying the airbag without accelerometer input would actively harm the driver.

Once when I was on a walk, a car activated all airbags just before it passed me while being driven in the opposite direction. I was more or less looking at it all the time, so there was no external factors - just a very loud thud.

The speed was perhaps 25 km/h which was good, because the young lady in the car just kept driving straight ahead - no braking or steering or anything - until the car slowly hit the kerb and then a sign post 75 meters down the road and came to a stop.

There was no external damage to the car - or perhaps the fender was slightly dinged.

I walked there, while figuring out what had really happened and noticed that the driver was just sitting there in the car while it started to be filled with white smoke.

When I got there the car was more or less full of smoke, and the young lady still in the drivers seat looking straight ahead and appeared to be very confused.

Someone else got there a few seconds before me and opened the doors to the now totally smoke filled car, and the lady sort of woke up - and she got out just as I got there.

I think she was physically fine - no arms broken although that could happen when an airbag deploys - but she was really confused although she was starting to get a grip after a minute or so, and started to worry about the blown up interior in her car...

I guess it's one of those things you really don't know how you will react - she totally froze.

If that would have happened at any sort of speed, I'm pretty sure it could been a fatal accident.

Why did the car activate the airbags? What car is this? I've never heard of airbags deploying improperly before (except when we have a busted set that we use to launch a tire a million feet into the air).

I don't remember what car it was, it was somewhere between 7-10 years ago. Possibly a small Toyota, or another asian car.

I remember that I had to look it up at the time - and apparently a spontaneous deployment is not unheard of, but rather uncommon. Old news article: http://www.autonews.com/article/19980824/ANA/808240708/airba...

> I guess it's one of those things you really don't know how you will react

In my case of the hypothetical close-call, I think if I had a car with one of these "impending doom" tones, just hearing that tone in conjunction with a near-miss on the road would probably give me a heart attack.

I read "It takes just 150 milliseconds to deploy the reversible seat belt tensioners" as 150ms being the time to fully tension them (and then, possibly, for the case where they have to be tensioned most, for example when a passenger is leaning forwards), so some effect will be visible before that.

And useless? There certainly are situations where this will help, for instance when the driver is braking hard. Also, if you hit a wall head-on, a Mercedes will have over a meter of crumple zone. Pulling the passenger away from the car interior by the time that crumple zone has done its job will surely help, even if it only a few centimeters.

Seat belts also should not crush the passenger to death. That puts limits on the speed at which they can be tensioned (it wouldn't even surprise me if the weight of the passenger would be used to adjust the pretensioning force)

Yeah, I wonder if that's a mistake on the webpage. Some other site I found[0] says the decision to deploy tensioners happen at 0.015s after impact, and they "go into action" at 0.020s. Someone might have accidentally added a zero when converting units?

[0] http://www.takata.com/en/around/seatbelt01.html

No, they lock your seatbelt before you move. Modern ones do this during heavy braking: all you notice is that when you step on the brake your belt is already locked.

That's not the pretensioner. The pretensioner is a one-shot pyrotechnic device (i.e. explosive) that pulls in the seatbelt. It's only activated if the car senses a crash is happening.

It's an ambiguous term. You're talking about a pyrotechnic system that can only be used once, but there are also electronic systems that are a part of daily driving. My point is that the 150ms happens before the person moves at all, so 30mph is not part of the equation.

I defeat this by driving with my left shoulder far forward.

This is because it nearly killed me several times. When in a dangerous situation, I may need to brake and/or turn my body to look behind me. It's the "and" which is trouble. If I brake and then need to look, the shoulder belt might stop me, leading to a crash.

The fundamental problem is that restraints are designed for 100% passive crash dummies. Restraints are not designed for people who must actively control the vehicle.

Reply on news.ycombinator.com