nav search
Data Centre Software Security DevOps Business Personal Tech Science Emergent Tech Bootnotes
BOFH
Lectures

back to article
Airbag bug forces GM to recall 4.3m vehicles – but eh, how about those self-driving cars, huh?

Silver badge

Checking online might be difficult if...

you were this person...

"The issue has reportedly already been blamed for one fatal crash"

I'm sure the lawyers have it covered, move along, nothing to see here.

8
0
Silver badge

Re: Checking online might be difficult if...

That's very true. But if you don't want to be the second person, get it checked. With exception of one or two models, most seem to be "high end" which strikes me as strange. One would think the high end vehicles would get the "better" systems.

3
0
Silver badge

Re: Checking online might be difficult if...

The "better" systems are probably more complicated, with a bigger surface for bugs.

There's good things to be said for cheap'n'cheerful.

25
0
Anonymous Coward

Re: There's good things to be said for cheap'n'cheerful.

And even better things to be said for lean'n'mean at the right price.

Not to be confused with Lean as once applied to manufacturing processes and now spreading like Japanese knotweed to places where it really doesn't belong and is desperately difficult to exterminate if it does get a hold.

8
0
Silver badge
Holmes

Re: Checking online might be difficult if...

"One would think the high end vehicles would get the "better" systems."

But not the better drivers, eh? Think about it.

2
0
Anonymous Coward

Insider's View

GM airbags have had lousy code since 1994. In some cases they even buy a microcontroller that is so cheap, IT DOESN'T HAVE A WATCHDOG!!!

The US National Highway Traffic Safety Administration (NHTSA) says that rather than deploy the airbag, the software will instead default to a test state.

Bad, bad, lousy code being written in Kokomo.

0
0
Bronze badge

Re: Checking online might be difficult if...

Actually the cheaper vehicles see more investment in their development as a general rule of thumb.

Simply because they sell more of them, the margin is smaller, but the total profit is higher.

0
0
Silver badge

Voluntary

"The recall is voluntary" - Why?

I would suggest that the fix should not be voluntary and the onus be on GM to go out and proactively fix something that presumably (*) violates mandated safety requirements.

(*) the small print: I am not an American and am making this up as I go along

13
0
Anonymous Coward

Re: Voluntary

Voluntary

"The recall is voluntary" - Why?

Mainly marketing. Making it mandatory would make it look worse, that's also why it's "in certain rare circumstances when a crash is preceded by a specific event impacting vehicle dynamics" instead of "after you have waxed you car and then hit anything more solid than a pigeon".

The problem is that the next death will thus be evaluated with a view on wether they stressed enough that it wasn't really an optional update, at which point that verbal marketing trick will become VERY costly and the recall will still have to be issued. I think it's a stupid, weaselly thing to do.

By the way, there are no "luxury" or "pleb" airbag variants that I know of :)

Last but not least I'm puzzled by the attempts to give this a self-driving spin. There's no material difference between the dynamics of an accident in either case. As a matter of fact, the befit of self driving cars is that they don't need an owner's or garage staff's time to get the car in for service. Nor will they mind waiting (well, whoever operates them might in the case of taxis).

3
5
Silver badge

Re: Voluntary

GM is issuing the recall before it will be mandated by the regulators. Also, this puts them in a more favorable civil liability position of they realized there was a problem and issued the recall in presumably a timely manner. So suing GM after the recall for this defect is not an option if the defect causes an accident or injury. GM can now argue, you were advised to fix the problem by us you did not thus it is solely on you. This narrows the potential window of legal liability and potential damages.

One of the reasons the Ford Pinto was notorious was the fact Ford apparently knew of the problem but did nothing for several years.

2
0
Silver badge
Devil

Re: Voluntary

Last but not least I'm puzzled by the attempts to give this a self-driving spin. There's no material difference between the dynamics of an accident in either case.

The remark concerned the presence of similar bugs in any module of a self-driving car's software: some rare event X happens, then event Y happens which the car can't react to because event X hasn't been handled properly and has put the module into lalalacan'thearyou mode.

Nor will they mind waiting (well, whoever operates them might in the case of taxis).

Customer has just boarded a self-driving taxi and ordered it to proceed to the airport at speeds in excess of R17 because there's the only plane to Outer Elbonia for the next two weeks to catch. Then shortly into the trip the garage signals there's a free slot for the five-hour software and hardware update coming up in five minutes. The glitch has just happened and the car has erased all instructions by the passenger, and even that there's a passenger at all. Of course the doors are locked, as that's the default when anywhere else than the pickup and dropoff location.

3
1
Silver badge

Re: Voluntary

For the consumer nearly all recalls are voluntary in that if you don't make an appointment to get it fixed they aren't going to come to your house and force the repair on you. The voluntary part here is that GM voluntarily performed the recall as opposed to having the recall the vehicles mandated by the NHTSA.

0
0

Process is far too slow

Speaking as a Honda owner who has just had one of my front airbags replaced under a recall notice... these things need to be done quicker as it took year from buying the vehicle to getting the recall notice... and the recall notice had been out there for a year, I was already aware of it and was told by Honda to 'wait for the recall letter'... which arrived in August after I told them I'd just purchased the car in July 2015 which was already after the recall had been issued.

Voluntary isn't good enough, these kinds of things need to be mandatory and notices issued to all registered owners.

But as it's all about money... it would have taken the risk of losing more money through potential lawsuits than the cost of the recall to actually make any company take action.

15
0
cd

Re: Process is far too slow

Honda recalled the driver's side AB. I asked about the passenger side when I took it in, "no recall on that". Now, a few months later, there is. The car is 12 years old, seems to me they already knew but were delaying, hoping to save money.

If elected I will build a machine that turns greedy upper management into dog food and will use it liberally.

16
0
Silver badge

Re: Process is far too slow

What did dogs do to you ? Nope grind them up and feed them to each other.

16
0
Silver badge
Coat

Re: Process is far too slow

"t would have taken the risk of losing more money through potential lawsuits than the cost of the recall to actually make any company take action"

I do believe that is referred to as "The Pinto point" in the auto industry.

11
0
Silver badge

Re: Process is far too slow

See Fight Club, Edward Norton explaining "The Formula".

2
0
Anonymous Coward

Re: Process is far too slow

What did dogs do to you ? Nope grind them up and feed them to each other.

I don't think they'll eat much after being ground up, or am I being too logical here?

:)

7
0
Silver badge

Re: Process is far too slow

I've had a Honda recall notice too on an 8 year old car. I presume you guys are in the US as over here they've sent 2 letters now advising of the recall but in both stated they haven't got enough parts to have the cars in to fix. This recall is not a software bug though it's an over explosive airbag that might cause more damage than it prevents. Software bug or not, I think any safety equipment that hasn't been thoroughly tested for faults is unforgivable.

4
0
Silver badge

Re: Process is far too slow

The airbag issue was not specifically Honda but caused by the airbag manufacturer. These bags were supplied to many automakers so the replacement parts may be in short supply given the total number of cars and number of bags to be replaced.

2
0
Silver badge

Re: Process is far too slow @Alistair

It was called the pinto problem to us when we studied Mech Eng, interestingly enough Chrysler were getting investigated in 2012, for design in Grand Cherokees that has the petrol tank at the rear, above the axle, using cheap materials. Which I guess goes to show not every engineer was taught about it, or maybe there really or some gits out there running car companies. I remember seeing a program showing the tank location (that wasn't about the investigation), and the first thing that came to mind was "um Ford Pinto".

2
0
Silver badge
Mushroom

Re: Process is far too slow

Nope grind them up and feed them to each other.

In the end there will be one C*O left (if you grind them all up there will be none to eat the stuff), and he will have been eating greedy C*O muck.

You need to think of a safe way to dispose of that one.

0
0
Silver badge

Re: Process is far too slow

Well only grind up 50% them then.

0
0
Silver badge

Re: Process is far too slow

Hydrochloric acid is you friend. I see your point. I'll leave 20 C*O behind to do the eating.

0
0
Meh

Re: Process is far too slow

these things need to be done quicker as it took year from buying the vehicle to getting the recall notice

To be fair, Honda probably did not have several million spare airbags sitting in a warehouse somewhere. Nor did they have millions of service technicians qualified to install all of them at once.

So they did it in stages/groups of however many they thought they could reasonably manage at once while not making it take too long for any one group.

0
0
Silver badge

Bah!

Why is it that the more "smarts" that are built into working devices, the less reliable they become?

8
0
Silver badge

Re: Bah!

"Why is it that the more "smarts" that are built into working devices, the less reliable they become?" --- Stevie

Well, we know the answer to that --- the more code you have, the more bugs you have. What I don't understand is why the switch from test to armed has to be in software. Seems to me to be crying out for a hardware solution. It's not like a grenade pin --- the force / speed combination required to trigger a deployment is fairly substantial --- I'm not sure how realistic it is to subject a vehicle to a non-destructive test that would deploy the airbag once it is assembled.

Further info from someone who has a clue very welcome, I'm clearly missing something.

8
0
Silver badge

Re: Bah!

Ah, but if they were REAL smart cars then surely they could leave a message to the owners, along the lines of "Sorry, had to go for a refit, TTFN" and drive themselves back to GM?

5
0
Anonymous Coward

Re: Bah!

" the more code you have, the more bugs you have."

That's a start, but I'd prefer:

The more you rely on cheap computers (hw and sw), the more bugs you are likely to have.

"why the switch from test to armed has to be in software. Seems to me to be crying out for a hardware solution. It's not like a grenade pin --- the force / speed combination required to trigger a deployment is fairly substantial --- I'm not sure how realistic it is to subject a vehicle to a non-destructive test that would deploy the airbag once it is assembled."

Quite. Car manufacturers and inappropriate use of "test modes", eh.

That said, the last time I saw an "airbag test system" (what the manufacturer uses in car factories to record test status, VIN info, etc) it was admittedly a couple of decades ago, and it was more about testing the sensor and controller than testing the airbag itself.

Trouble is, hardware usually costs money, and software comes cheap these days (that's what manglement seem to think anyway).

7
0
Anonymous Coward

Re: Bah!

Ah, but if they were REAL smart cars then surely they could leave a message to the owners, along the lines of "Sorry, had to go for a refit, TTFN" and drive themselves back to GM?

Until my car learns how to swim* I'd prefer it didn't. I'm on the wrong side of the ocean :).

* that would make for a really, really smart car :).

0
0
Silver badge

Re: Bah!

Complex systems have more to go wrong, software or not.

The more moving parts, the more to go wrong.

The more components, the more to go wrong.

The more complex the parts, the more to go wrong.

This is why stuff used to just be built simple. It was often a conscious decision not to over-complicate things with feedback loops, electronic control, etc.

Nowadays, everything suffers from systemd syndrome. Your phone tries to do everything. You car tries to do everything. Your social media network tries to do everything. And they all do all those things, from all manufacturers, and are competing to be "the one" that does whatever. All that happens is that you get ten devices, none of which are specifically designed for the task at hand, that you only use a handful of features on each.

11
1
Ogi

Re: Bah!

This! Parent hit the nail on the head. Complexity breeds problems, the mark of a good engineer is to make something as simple as possible, but no simpler.

Problem is any idiot can make a stupidly complex system. It takes smarts to make an elegant simple system that does the same thing. Unfortunately idiots are cheaper to hire, so this is what we get.

Also there is always a push by accountants to save costs everywhere. Before, cars had separate computer systems. One for the engine, one for the alarm system, one for the air bags, and a separate entertainment system. If there were any interconnections, they were specific and targeted to components that required communication.

Now it is usually all handled by one powerful control unit with thousands of wires coming out of it, with generic addressable buses where everything hangs off. and it is possible that a bug somewhere overwrites the stack and instead of sending a signal to the engine it triggers the airbags.

As for me, I am going to stick to my 80s car thank you, although it is getting harder and harder to stay away from new cars due to changes in insurance policies (more and more are refusing classic car insurance unless you have a modern car as well).

Worst thing is, car manufacturers don't seem to care. They are making things more and more complicated, now with "AI" and "Constant connections" it sounds like a dystopian disaster in the making.

3
1
Silver badge
Coat

Re: Bah!

Ah, but if they were REAL smart cars

they would drive to a Mercedes dealer.

1
0
Silver badge
Devil

Re: Bah!

and it was more about testing the sensor and controller than testing the airbag itself.

As the actual airbag is some kind of electrically triggered explosive which you can test separately without any car parts around it, it's the sensors and controller software that matter.

Trouble is, hardware usually costs money, and software comes cheap these days (that's what manglement seem to think anyway).

A hardware solution costs umpteen microrupees extra per unit, so that adds to the cost for each car produced. Software costs umpteen-and-a-bit rupees up front, so you'll break even at a million-and-a-bit cars; beyond that it's Pure Profit!

1
0
Bronze badge
Thumb Down

Re: Bah!

Ogi, for 99% of vehicles they are still multiple controllers, typically running over a CAN bus.

Mainly due to weight, a controller is usually lighter than that massive bundle of wires and from the OEM's perspective easier to build than adding 200 miles of wiring.

The main priority after getting a vehicle to "work", is making it cheap to assemble.

0
0
Silver badge
Pint

And the 'Sentence of the Year' award goes to...

El Reg "Fingers crossed there are no programming cockups like this one when fabled self-driving cars arrive, eh?"

Congratulations.

Bonus points for the word "fabled". Nice touch.

8
0
Silver badge
Thumb Up

Re: And the 'Sentence of the Year' award goes to...

"Nowadays, everything suffers from systemd syndrome. "

I prefer this. With big brass knobs on.

The epitaph of the early 2000s.

5
1
Silver badge
Terminator

Re: And the 'Sentence of the Year' award goes to...

Well, seriously. Mentioning "bug" and "self-driving car" in the same sentence is kinda stupid though. If it is "self-driving", does the concept of "bug" still apply? Cause maybe you can't even iron it out, the problem will be hidden behind walls of machine learning results that no-one understands. It will be so complex that we will have to talk about "inappropriate decision-making" or "cognitive excursions" etc.

Same as with human drivers, actually.

4
0
Silver badge

"already been blamed for one fatal crash and three others involving serious injuries"

Hmm, how can you blame a crash on an airbag which isn't supposed to do anything till a crash is actually occurring and didn't?

As it has taken a couple of years for the problem to come to light it would seem it is rare to enter this diagnostic mode or it is in that mode for a small proportion of the time.

I have been driving for 40 years without an air bag deployment or something that would have deployed an air bag in cars that didn't have them. I wouldn't care if my car didn't have an air bag at all, never mind one that is just a bit unreliable deploying.

3
3
Silver badge

Re: "already been blamed for one fatal crash and three others involving serious injuries"

I think they are referring to the crash being unexpectedly fatal. It's very easy to die in even quite moderate shunts without an air bag (and seat belts).

I can see it taking a couple of years, easy. Unless the dev team is in place, equipped with their original build chain, dev environment and laboratory, it'll take ages simply to get to the point of being able to do anything. And then they have to try and recreate the crash conditions, possibly involving crashing actual cars fully equipped with a substantial diagnostics suite, in the hope of seeing the software failure in action (bearing in mind that it presumably hadn't failed in whatever tests they were doing back when the software was originally developed).

Ford had airbags going off spontaneously years ago, happened to a friend. That's pretty dangerous, fortunately they kept control and made it to the side of the motorway. Ford UK's responses were un-reassuring...

6
1
Silver badge
Coat

Re: "already been blamed for one fatal crash and three others involving serious injuries"

Well, I was wondering... if it were prone to exploding at random times, then yes, it should be fixed ASAP, but an airbag that doesn't go off is just the same as not having one.

1) the airbag was introduced because large populations of drivers were apparently unable to use an existing well-tested solution: the seat belt.

2) airbags go off in a front impact. How often have you had one of those? How often have your airbags gone off? If someone drives into you, they're almost certainly going to do it at a combined speed which would render the passenger cell useless anyway, and if you drive into someone else you should have been paying better attention.

A technological solution for idiocy? It merely allows idiots to breed better idiots!

--> the flameproof one...

5
5
Anonymous Coward

Re: "already been blamed for one fatal crash and three others involving serious injuries"

"1) the airbag was introduced because large populations of drivers were apparently unable to use an existing well-tested solution: the seat belt."

No they were not, air bags should not go off if you do not have a seat belt on, they will cause greater injury, as in death, if you don't have a seat belt on. They are to reduce neck and head injury caused when wearing a seat belt.

7
2

Re: "already been blamed for one fatal crash and three others involving serious injuries"

2) ....How often have you had one of those?

Once in 30 years of driving.

2 contd.)...How often have your airbags gone off?

Once

2)If someone drives into you, they're almost certainly going to do it at a combined speed which would render the passenger cell useless anyway

I walked away with bruising and mushy brains that lasted for 6 months... the other not so much; she was in hospital for a fortnight. Each travelling at approx 60 mph, 30% front overlap head on. Passenger cell worked in both vehicles - what speeds were you thinking of?

Without doubt car manufacturers are afflicted by the cost of death - when is the cost of fixing a problem outweighed by the cost of potential litigation... same as software really.

0
0

Re: "already been blamed for one fatal crash and three others involving serious injuries"

The Takata airbag issue is not random inflation but spontaneous and explosive detonation of the propellant filling the drivers chest with hot metal shrapnel

5
0
Silver badge

Re: "already been blamed for one fatal crash and three others involving serious injuries"

@AC,

"No they were not, air bags should not go off if you do not have a seat belt on, they will cause greater injury, as in death, if you don't have a seat belt on. They are to reduce neck and head injury caused when wearing a seat belt."

Er, it depends on where you are.

Across Europe and vast swathes of the civilised world it's normal (and a legal requirement) to wear a seat belt. The thing that actually saves your life is the explosive charge in the seatbelt pre-tensioner that goes off as the shunt starts. This hauls you back rigidly into your seat, and allows the crumple zones of the car to absorb your kinetic energy as the passenger cell (and the now firmly restrained passengers) decelerates. The airbag (which is actually quite small) simply goes off to prevent you face being mushed by the steering wheel or dashboard and to take some of the energy out of the head, reducing the whiplash caused by the body recoiling back into the seat (where the headrest takes out the remainder).

Without the belt and pre-tensioner you're simply a 40+mph missile moving forward through the passenger cell, not being slowed down by the crumple zones at all, injuring or killing everyone else you hit along the way.

In the USA, where weirdly there's states where you don't have to wear a seat belt, the airbag takes on the primary job of decelerating the passengers. Consequently they have to be much bigger, needing to fill the space between the entirety of the passenger's body and the dashboard / steering wheel. They go off as the shunt starts, aiming to cushion the blow as the passenger's unrestrained body hurtles inexorably towards the dashboard and steering wheel. Clearly this is an inferior solution, is of no help whatsoever in a side-swipe, and actually leads to people not sat normally in their seat (leaning forward, for example) being killed by the airbags going off in quite minor shunts. An airbag is quite capable of decapitating someone. Nasty.

Why Is it Law To Wear A Seat Belt

This is because an unrestrained body inside a car involved in a high speed shunt can be lethal to other passengers in a car. For example, if you're a passenger in the back seat not wearing a seat belt, you will likely kill the driver in the front if the car is in a fast enough shunt. It's also not inconceivable that an unbelted passenger in one car kills people in the other in a head on collision as they fly through two windscreens. I turf out passengers who refuse to wear their belt; I don't want their stupidity to kill me if some other idiot decides to drive into my car at high speed.

The extra cost of your lengthier rehabilitation courtesy of the NHS subsequent to your more severe injuries in the less likely circumstances of you surviving such a crash is a secondary consideration.

This is also related to the reason why air passengers are told to put things away during landing and take off and put their seatbelts on. Things like laptops, phones, books, etc. all become 150mph+ missiles hurtling through the cabin if the plane crashes. Your ability to get out of a burning aircraft is greatly enhanced if you've not been knocked unconscious by some other idiots laptop. Same reason why they're not keen for people to be standing up during taxiing (might be doing 40mph+), or on take-off / climb or landing / approach; passengers have no warning of how the aircraft will manoeuvre / collide next, and unseated / unbelted they're 80+kg of fuckwit selfish lard just waiting to become a deadly flying object within the cabin. You only have to look at the carnage caused by high altitude turbulence where unbelted passengers cause a lot of injuries to others as well as themselves.

In all transport an unbelted passenger is a prime example of an ignorant and selfish human being. It's notable that the only forms of transport where belts are not generally compulsory is ones that are slow (city buses, trams), or not normally exposed to rapid decelerations (trains, ships, submarines). Of course when a bus or train or ship does have a high speed crash the carnage is immense. For example a sailor unfortunately died of head injuries and 98 were injured when USS San Francisco collided head on with a sea mount at 30+mph.

Trains operate at high speed but in highly constrained environments (the rails), and generally the only thing that is a threat to them is another train, wheel failure or derailment. But even then modern designs are quite remarkable. The seating is designed to help keep passengers more or less in place should a train derail. They're also pretty good nowadays at not coming apart in crashes, saving a lot of life. However, you're still probably better off in a backwards facing seat in the middle of a carriage in the middle of the train.

9
0
Silver badge

Re: "already been blamed for one fatal crash and three others involving serious injuries"

I have been driving for 40 years without an air bag deployment or something that would have deployed an air bag in cars that didn't have them. I wouldn't care if my car didn't have an air bag at all, never mind one that is just a bit unreliable deploying.

A few years back a friend of mine had an accident on a country road when a speeding driver came round a corner right into them. There was an offduty police office behind my friend who estimated the impact at near 100mph. He got out of it with a sprained ankle and some bruising, (when the fire crews and scrappy said that usually when you see a car in the state his was it would be the sort of thing they would expect to be a fatality). He has two cars if he had been driving the one that was five years older or so it could well have been.

There's a reason they keep improving things like airbags and crumple zones, it's to increase survivability. I believe it was top gear yonks ago that tried crashing an older model 5 star NCAP rated espace against one of the newer ones (by a similar amount of years). The new one pretty much ploughed through the older mode on a front corner impact test, the people in the front would almost certainly have been dead.

0
0
Anonymous Coward

Re: "already been blamed for one fatal crash and three others involving serious injuries"

exactly my thoughts, most head on collisions with another car will be at speeds form 60mph up to 120mph (each car doing between 30 and 60)

So surely cars should be designed to ensure suitability in a 120mph crash...

0
0
Silver badge

The converse situation

The cited circumstances are diagnostics interfering with the deployment of airbags.

If this is possible then I can foresee circumstances occurring where diagnostics interfere when deployment is emphatically not required.

"Someone I know" caused a domino effect on the motorway one day when he went through a particular sequence of moves which provoked an irrational braking event from the car management system. He didn't quite make the traffic news partly because (thankfully) nobody was killed.

So if airbags were to deploy when they shouldn't sounds to me like a serious matter. One would hope exhaustive testing is being carried out in this direction too.

3
0
Anonymous Coward

The recall does not extend to Bulgaria

... since they have their own equipment in place there.

11
0
Silver badge
Pint

Re: The recall does not extend to Bulgaria

Sir!

0
0

Page:

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

The Register - Independent news and views for the tech community. Part of Situation Publishing