Usually, we don’t report on recalls. If we would, we would be reporting on recalls all day. Sometimes, we make an exception, if the matter has wider ranging implications. This is one of those times.
The recall: BMW recalls “approximately 50” BMW X3. They have electric power steering. There is nothing wrong with the power steering. But there is everything wrong with the computer that runs the power steering. The steering has a position sensor, and the sensor’s calibration data pertaining to temperature balancing was mistakenly overwritten.” At some point, the computer finds out that something is wrong, and shuts off the electric power steering (EPS).
The wider ranging implications:DARPA challenges colleges to build autonomous vehicles. Google already has “look, ma, no hands” cars drive autonomously through the city, now they want some that read your mind. Cars drove autonomously from Italy to China. Nevada paves the legislative roads for driverless vehicles. What do they all have in common?
Electric power steering. Along with all kinds of other computer gadgetry than can go wrong. This is definitely a topic worthy of LaHood’s attention. The entire concept is only as robust as the “weakest-link” in the car’s trinity of mechanical/electrical-electronic/software systems. It relies on the 1st tier supplier’s ability to manage “pharma-level” quality from it’s sub-tiers, or to conduct “NASA-level” fault detection prior to delivery to the OEM (which BMW’s German-Swiss steering supplier clearly failed to do).
Long-proven complex systems and new simple things fail because somebody screwed-up or cut-corners, or both. Car companies and their supply base have proven time and again, for the last 100 years, that they are not able to achieve the pharma-NASA level, maybe because it does not really exist. Even if it does, it did not spare NASA from the occasional screw-up.
If the pinnacle of product development and manufacturing goodness promised by the nexus of premium-price German technical excellence (BMW) and Swiss quality (ThyssenKrupp Presta Steering) can’t protect itself from the stupidity of overwriting a block of essential data, imagine the mayhem when all the “lesser” (or greener) technology companies try to climb aboard the autonomous bandwagon. If the EPS is shut off, the human behind the wheel will have to work a bit harder, but he or she will still be able to steer.
Without a human behind the wheel … I know, the car will receive a recall notice drive, and dive autonomously to the dealer to have itself fixed.
Long-proven complex systems and new simple things fail because somebody screwed-up or cut-corners, or both. Car companies and their supply base have proven time and again, for the last 100 years, that they are not able to achieve the pharma-NASA level, maybe because it does not really exist.
It’s doesn’t have to be perfect, just better than most human drivers. Which, at least in the US, isn’t really such a high bar.
Also, you should note the rise in commercial aircraft automation has resulted in a precipitous decline in accidents.
Bertel is not used to American drivers, who do not pay attention even when they are awake and sober. Me, I figure that even malfunctioning computers are better than some of the drivers I see around here — texting, while their free hand holds a drink, and their little dog is charging around the cabin like a superball on meth.
Ideally, we would all drive the same car.
Yeah, but are you going to get it back to me in time for me to go the store tomorrow afternoon?
The entire concept is only as robust as the ”weakest-link” in the car’s trinity of mechanical/electrical-electronic/software systems
What, you mean “software” as in the nut behind the wheel?
Most accidents are caused in whole or in part by human error, and modern, complex technology being decried here actually makes cars safer because a) passive and active safety systems work, and b) you can have things like, eg, multiple redundant sanity-checked sensors, rather than a single mechanical cable. Given the number of hours driven by everyone, everywhere, mechanical/electronic/logical errors are few, far between, and more often than not issues of maintenance anyway.
Much of the brouhaha around Toyota was due to people who couldn’t sort out the pedals in a panic, and the most publicized instance—the CHP officer—was caused by an ostensibly-trained driver who couldn’t figure out how to shift a car into neutral.
I’ll take robot cars eight days a week, thanks.
Although I have a different take on the Toyota scandal, I have to agree with you that the electronic-control deniers are mostly from the “Bring back carbs and drum brakes!” set of automotive reactionaries.
OK, I’ll bite. Because I have been an electronic control denier since about 1989, when some Chevy dealer service manager told me my Celebrity wasn’t running because the crank position sensor was fried. Wait, what happened to the cam, the crank and the distributor all being mechanically connected? I do understand, even if I don’t like it, how electronics improve engine efficiency. But how in the Sam Hill does “electric power steering,” whatever exactly that is, dependent on some gizmo to tell some other gizmo where the wheels are, do anything other than create more expensive crap to break? Power losses from a modern conventional belt driven P/S pump aren’t that great. And the electricity to run an electrically powered P/S pump isn’t free in any event.
“Human error.”
Unless you know something I don’t, all the nifty systems you described are designed, built, and tested (or not tested!) by … Humans.
“If the pinnacle of product development and manufacturing goodness promised by the nexus of premium-price German technical excellence (BMW)”
I guess that people are still falling for that “German engineering” marketing nonsense in droves. Even after the HPFP failures, the plastic impellers, the nonsensical wiring, the $60 oil-filter cup removal tools, the imploding window regulators, and the whole litany of other sins committed by BMW and the other German automakers. VAG’s continued cynical attempts to milk this myth are only going to get more comical, as their crop of nasty new discount specials age and people see just how bad things can really get. They must have P.T. Barnum as their CEO by now to keep the charade going.
I don’t think the issue is “German engineering” but just more multi-national corporations focusing on sales and marketing and stock prices ahead of any other considerations.
That is not to suggest that German engineering is necessarily any better than any other but it surely is no worse.
As I’ve said, the problem is bean counting.
Today’s Wall Street Journal:
Q: I have a 1993 SL 600 with 35,000 miles on it. A dash light has come on and the dealer informs me that I need a new wiring harness. This will cost over $10,000. They told me that for several years that Bosch made a biodegradable harness and that Mercedes did not stand behind the failure. …
A: Many Mercedes-Benz models cars from 1993 through 1995 had “biodegradable” insulation on their wiring harnesses. The insulation tends to crack and crumble after a decade or so regardless of overall mileage or how often a car is driven. … I think you can shop around and find one for less. … Mercedes-Benz … may offer a discount or another way of defraying the cost.
I thought that was to comply with some sort of green initiative.. a percentage of the car had to be biodegradable or recycled or some such mumbojumbo.. At least that’s what I was told years ago.
It’s too bad that we have apparently long forgotten the KISS rule, especially on safety-critical vehicle systems such as throttle, brakes & steering.
Example: my (now-departed) 1988 Buick Electra T-Type came equipped with antilock brakes that used an electric pump to provide power assist and rear (drum) brake system pressure. The system worked great, up until the pressure switch controlling the pump stuck on, causing the pump motor to stay on and burn up the pump, resulting in a sudden loss of power assist and all rear brakes.
Now, on paper, you still have full manual front brakes so what’s the big deal? Your front brakes do most of the work, right? Well, after the failure I drove it home in 2nd gear using neighborhood streets at not over 20mph, and it was still scary due to the extreme lack of stopping power. Thankfully, the failure occurred as I was inching forward into a parking spot, and still I almost ran into the building ahead of me. If it has happened out in heavy traffic I most likely would have rear-ended somebody in spectacular fashion.
So, the then-optional “safety” system added to my vehicle, which only engaged at most a half-dozen times in 16 years and over 150K miles of driving, could have in itself caused a major accident when it failed. Did having antilock brakes help me to avoid an accident? Maybe one time for certain (would have kissed the side of a BMW at 5mph on glare ice).
So, is the added complexity and it’s accompanying increase in potential failure modes worth the benefits of said system? That’s an argument we will continue to have.
Well, you did buy a turbocharged car, which was the main reason for the complexity.
If the throttle was stuck open, the engine would be making boost, meaning there would be no vacuum available to operate a conventional vacuum-assist braking system.
The ABS was not the main reason for the electric motor.
There are other ways around that problem for turbocharged or supercharged cars. “Hydroboost” systems work off pressure provided by the power steering pump.
Finally, I am sorry that a failure put you in a dangerous situation, but that pump didn’t fail in an instant. If you were paying enough attention, you would have heard the pump motor running at inappropriate times and could have avoided the failure. (If you are hard of hearing, then never mind.)
On the other hand, I am sure GM never informed you to watch out for random humming from the engine bay.
Nope, no turbo – the Electra T-Type was only available with the 3.8 or 3800 normally-aspirated motor.
The pump was very quiet, barely audible even with the engine off (due to excellent rubber isolation bushings), so it would have been difficult to detect. I contemplated adding a LED to the dash to show when the pump was running afterwards but never implemented the idea.
Oh, an ELECTRA T-Type. My mistake!
redmond: There are some individuals at EPAS suppiers or their OEM customers that are robbed of their REM sleep by the possibility of “system self steering.”
In the 1st generation Z4, BMW used electric power steering for the 2.5 and 3.0 liter roadsters.
The model that wasn’t equipped with EPS? The range topping BMW Z4M roadster and coupe.
I can only surmise that BMW thought that EPS produced steering feel that was unacceptable for buyers of the Z4M, but would suit the punters buying 2.5 and 3.0 liter cars.
Oh, if ONLY the suppliers can reach “pharma-level” quality assurance, then it will be just great.
http://money.cnn.com/2010/05/14/news/companies/tylenol_recall_plant_conditions/index.htm
And then they will reach “NASA-level” quality testing. Awesome.
http://www.lompocrecord.com/news/local/military/vandenberg/article_6b8b3be8-46f8-11e0-89ea-001cc4c03286.html
http://www.techxilla.com/2010/05/01/nasa-great-failures-of-history/
http://www.spacedaily.com/news/oped-04zh.html
Part of making autonomous operations failsafe will include on-board self-diagnostics that will make OBDII seem as primitive as cave drawings, coupled with constant cell phone communication to a central computer to ensure that all software patches have been applied. The downside will be that these vehicles will refuse to operate (at least autonomously) until all systems approve operation, and will shut down operation immediately if an error is detected. It will, however, most likely have the decency to also contact OnStar, AAA etc. on your behalf.
I hate to sound like some kind of tinfoil hat Luddite, but am I the only one who sees a potential problem with autonomous computer control combined with the inevitable addition of internet access in cars?
Is there anyone who will guarantee that someone out there cannot develop a virus that could infect the car’s computer via Twitter or some other route?
I hope someone really gives this some thought.
Someone has: check out FAA DO-178. Automotive applications like this one would call for Level A certification. It is neither easy nor cheap to do such a certification but many many companies have (SAGEM, Rockwell-Collins, Bendix-King) and there is an industry that already exists to assist in the verification and validation (V&V) required to move this along.
@Jet_silver: guidelines and certifications for software are a start, but a big problem is the fact that software development is not as mature as other Engineering disciplines. We’ve only been developing software for a bit over 50 years, compare that with the length of time that we have been building bridges, for example.
There was an interesting quote (can’t remember the source) about the current state of Software development to the effect of:
“If we built bridges the way we develop software, we would have creative artists build the bridges, and then get unskilled labourers to attempt to keep them standing while traffic was running over them”
It will take time for software development to mature to the same extent as the other disciplines, and unfortunately there may well be a few software “Tacoma Narrows” type incidents along the way…
The current bio-automated drivers make plenty of mistakes, a lot more so than do the electronics of their vehicles.
Fully automated vehicles don’t have to be perfect to be desirable, they need only be statistically better than the human driven machines we have today. Looked at that way, it doesn’t seem like such a steep road to climb.
“If the pinnacle of product development and manufacturing goodness promised by the nexus of premium-price German technical excellence (BMW) …. ”
Surely you jest. There isn’t a single German automotive company which can claim any sort of leadership role in developing and deploying robust, high reliability automotive electronics. If you want reliable electronics, you are much better off in an Infiniti than you are in a BMW. For some reason, Lucas gets all of the bad electronics jokes, but the Germans have put out more dubious wiring harnesses, short lived anti lock brake controllers, and drive you crazy interfaces (iDrive anyone!) than has the rest of the industry put together.
Fully automated vehicles don’t have to be perfect to be desirable, they need only be statistically better than the human driven machines we have today. Looked at that way, it doesn’t seem like such a steep road to climb.
Really? I’ll think about that next time I shut off a Lincoln to reboot a navigation system, or a check engine light comes on in a Mazda, or an airbag warning light flickers in a Kia, or I have to take the battery out of my Blackberry, or my neighbor asks me to help with his Mac, or I hear a nightmare about an ABS failure, or the thermometer in one of my cars tells me a value I could determine was wrong merely by observing that water was liquid, or my wireless network signal strength varies as if someone added a lead barrier between my office and the living room, or a solid state drive proves to not live up to the hype, or when I hear about automated Metro crashes in DC(they have to figure out when to go and when to stop, but I’m sure steering and a lack of rails will make things more reliable), or pretty much any time I see someone frowning at a malfunctioning new device. People would all be dead if they had the lapses my cell phones have all exhibited. Maybe I have more faith in my fellow man than you do, but I know I have more faith in my ability to evade the mistakes of others than I do in an electronic device to reliably control my car. Sure, I’ve flown in fly-by-wire, computer stabilized aircraft. They aren’t made in GM numbers at GM prices, and sometimes they vanish anyway.
AF 447 is a perfect case-study.
We would do well to remember that there is a tremendous difference between fly-by-wire and steer-/brake-by-wire with the former representing aircraft-style thinking and execution (i.e. triple-redundant circuits) and the latter having automotive-style thinking and execution.
Maybe I have more faith in my fellow man than you do, but I know I have more faith in my ability to evade the mistakes of others than I do in an electronic device to reliably control my car.
Good point. That ability to evade the mistakes of others is an important element that’s lacking in current autonomous vehicle technology.
I’ve co-designed 2 aircraft collision avoidance systems and I’m currently working on core technology to solve some of the issues with autonomous vehicles. There is a certain amount of knowledge that humans possess as they gain driving experience that helps us predict the behavior of humans, animals, and other stuff that moves around on the roads.
Let’s say you’re in a vehicle driving down a down a road and a ball rolls across it. An autonomous vehicle would calculate the by the time it reaches the section of road where the ball crossed, the ball would have passed and there was no collision danger. As a human (if we have sufficient driving experience) we’ll realize there is a chance someone might be chasing after the ball and there is the potential of a collision with something that we can’t currently detect, so we’ll slow down or stop.
Needless to say, there’s a lot of work to be done (and a patent portfolio to be built ;^)).
John never claimed that all technology can work perfectly all of the time. He simply stated that at some point in the future, autonomous technology would probably progress to the level that it would be considered suitable for real-world driving applications. If you consider how far robotics in general have progressed in the past 30 years, there is nothing wrong with that statement.
” … I have more faith in my ability to evade the mistakes of others than I do in an electronic device to reliably control my car.”
In general, it has been well established that people FEEL safer when they have a greater sense of person control. But, that does not mean they actually are. Many people feel safer driving their own vehicle than they do as passengers in a commercial jet, but the statistics say that they are, in fact, much less likely to die in said commercial jet than they are driving themselves.
I’m not saying that autonomous cars are ready for prime time or that all technology is great. I’m simply saying that even the best drivers sometimes are distracted, sleepy or off their game … and that the poor drivers are out in droves all the time. Most “accidents” are actually avoidable screw ups by one or more involved people. Over 30,000 people are killed in the US every year thanks to said screw ups.
Part of google’s ted talk shows a kid running in front of the car — unexpectedly — and the computer slams on the brakes faster than any human could.
The computer has 360-degree vision, is never distracted, and can see problems and react with evasive maneuvers in tens of milliseconds. That’s a pretty big advantage over humans already.
Contrasted against Volvo’s collision-avoidance demonstration, where the car launched itself into the back of a truck trailer in front of a crowd of invited journalists…
Point-being, is not that the idea is bad (actually, I’m about as far from a Luddite as I think is possible, so I am looking forward to the day when I don’t have to guide the car from point A-B…), it is that timing (i.e. maturity of the tech) is key…
It is too early in the technology cycle, for a state (that probably is really clueless about what they are authorizing) to give a road-going permission (I don’t know the details of the authorization, so I hope it is quite limited) for autonomous vehicles (unless they are still equipped with a driver-over-ride system)…
Oh, I’m sure there will be problems. And even after many years, you will probably be able to find edge-cases where the computer-piloted car does worse than the human-piloted car.
Computer-driven cars floundered for years, with DARPA challenges ending with no one finishing. But whatever hurdle was holding them back has been cleared and we’re learning how to make them.
I don’t think states should be allowing driver-less cars right now — meaning a car where a human is not in the driver’s seat. But I think they should be allowing people to sit in the driver’s seat while a computer pilots. Let’s practice with that for a few years and we’ll probably be ready to let the computers take over.
I want one of these, but I really want my dad to have one of these.
Just last month BMW had to recall 900 or so X5 diesels for a much lower tech steering problem.
Summary:
BMW IS RECALLING CERTAIN MODEL YEAR 2011, X5 DIESEL SPORTS ACTIVITY VEHICLES (SAV) MANUFACTURED FROM MARCH 31, 2011, THROUGH MAY 19, 2011. DUE TO A COMPONENT MANUFACTURING ERROR AT THE SUPPLIER, THE WELDS IN THE BELT TENSIONER COULD HAVE INCIPIENT CRACKS. OVER TIME, THIS COULD LEAD A MALFUNCTION OF THE BELT TENSIONER.
Consequence:
IF THE BELT TENSIONER FAILS, THERE COULD BE A SUDDEN LOSS OF POWER STEERING ASSISTANCE, INCREASING THE RISK OF A CRASH.
http://www-odi.nhtsa.dot.gov/cars/problems/recalls/recallsummary.cfm?rcl_campaign_id=11V332000&prod_id=1210769&moduletype=VEHICLE&make=BMW&model=X5%20DIESEL%20SAV&veh_model_year=2011&searchtype=DrillDown
Meet the new tech, same as the old tech.
Re autonomous cars, If you don’t feel like driving, than take mass transit or call a cab.
Why the non-productive and arrogant input?
Robot-car had a crash…
http://www.nbcbayarea.com/blogs/press-here/Google-Blames-Human-for-Robot-Car-Crash-126845683.html