Slashdot Mirror
NASA Summoned To Fix Prius Problems
coondoggie writes "If you want to solve a major engineering mystery, why not bring in some of the world's best engineers? The US Department of Transportation's National Highway Traffic Safety Administration today said it was doing just that by bringing in NASA engineers with expertise in areas such as computer controlled electronic systems, electromagnetic interference, and software integrity to help tackle the issue of unintended vehicle acceleration in Toyotas. The NHTSA review of the electronic throttle control systems in Toyotas is to be completed by late summer." We're really in trouble when NASA has no choice but to call Bruce Willis.
So, this is an admission that sticking pedals and faulty floormats had nothing to do with the problem, and that the recalls to fix pedal and floormat "problems" were simply a smokescreen to hide the actual cause of the problem (albeit, unknown cause)?
Wait, so when a private corporation fubars something, you gotta roll in the government funded engineers to fix it? Interesting, interesting...
Ubi solitudinem faciunt, pacem appellant.
The budget cuts at NASA apparently keep them earthbound and working on earth crawlers
There's other theories, too, that NASA could help with. Such as current spikes or other hardware problems.
In reality, NASA may be a perfect choice given their experience with designing fault-tolerant systems. That means everything from protecting the system from the environment to software validation. The control systems in a car have become very complex, approaching that of airplanes and rockets. I think NASA is a good choice, although I might have gone with an aerospace company instead, such as Boeing, Lockheed, Northrop, EADS, etc.
Otherwise, it's just another set of computer scientists looking over a few million lines of code they didn't write, trying to find a defect that has supposedly manifest itself less than a few hundred times out of million of cars and probably billions of miles driven.
You're confusing "electronic" with "software." One possible theory is that interference (internal or external) is causing signals between parts to become corrupted. My understanding (having RTFA) is that they are focusing on the electrical engineering aspects of it. I would imagine that NASA, needing to design and test equipment in the harsh environment of space, is pretty darn good at exactly that.
Note to self: Stop putting jokes in my insightful comments so I can get something other than +1 Funny!
I said this earlier:
If NASA does find a problem then Toyota can spin it as it being so subtle that it took the resources of NASA to find it. They can then use this, with PR spin, and an agreement to contract with NASA for "consulting" as a win.
If NASA finds nothing, then Toyota is off the hook wrt the drive by wire system, again a win.
CAN Protocol(the de-facto automotive protocol) contains error checking. So if a node in the system sends out 00000001 but the "sun spot" turns it to 01000001, it finds that error. So unless it changes that to a 01000001, while also changing the parity bit(or whatever they use for error checking) to 0 as well(as compared to 1), WHILE ALSO somehow disabling the entire safety section of code that reduces the throttle input when brakes are applied... then I seriously doubt it.
Everyone involved in this knows the above facts, what they have to do is prove the above facts. The reason they called in NASA is because they lack the right type of experts, NASA does not. Case closed.
So basically, -1 troll/offtopic is really slashdots way of saying "I hate that you thought of something before me."
I'd really like to see how the computer in the car manages to consistently only enter this mysterious state when the driver is 60 or older (or maybe in the late 50s). Because normally, if you have a ton of examples of something failing, all of which involve people of an age famed for acquired inattentiveness or confusion, and which look just like many other reported and documented cases of elderly folks getting confused and hitting the gas pedal thinking it's the brakes, you'd not assume it was the computer.
My blog: http://www.seebs.net/log/ --- My iPhone/iPad app: http://www.seebs.net/seebsfrac/
Honestly, given the miles that Toyota has on land vs. the miles NASA has in space... Toyota might have the most 'accident free' record of the two. (Not including drivers who are just ignorant)
So the software producing and acting upon those CAN frames keeps the data in CAN frames internally, and has some sort of integrity check at EVERY layer? This could still be a hardware EMI problem. One that is usually caught by the CRC in each CAN frame, but in a few cases is affecting portions of the system that do not have an integrity check on them.
I think they are clearly software-related. But the question is what the root cause is.
You don't always know what the error condition is. It could be: "Well if the solder is cracked on this connection and this other sensor fails in this way, the following feedback loop is created and the software responds to the erroneous input by accelerating the car."
This is almost certainly some sort of cascade from the root problem. The cascade is software related. The root problem may well be hardware related. However I don't think you can just blame cosmic rays, etc. since the behavior described is too narrow. Why not uncommanded braking or power steering errors?
LedgerSMB: Open source Accounting/ERP
Mechanical throttles are conceptually simpler and have more obvious and obviously testable failure cases. It's not that hard to deal with them.
However, drive by wire introduces a number of layers of complexity and abstraction. In addition to mechanical failure cases, now you have electronic (hardware) and logical (software) problems as well. The added complexity makes all of this much more difficult to address, and it insulates the driver from the overall control of the vehicle.
This being said, I think it is obvious that where you have a hybrid car, purely mechanical acceleration is simply not an option. The complexity of the hybrid system itself requires electronic control, which more or less requires drive by wire of an equivalent. So this isn't an unconditional opposition. I just don;t think it is appropriate for traditional, gas-powered cars.
LedgerSMB: Open source Accounting/ERP
Why are we wasting American tax dollars to solve a foreign auto company's technical error!? To further drive American auto industries out of business? We should just ban Japan's defective lead-foot autos like we ban China's lead-filled products.
I wouldn't be surprised if this wasn't coming from NASA's already diminished budget.
That said, why is it in these stories of runaway acceleration, that nobody slaps the thing into neutral and hits the brakes? The stories always read like "I was powerless to stop my deathcar!" but drivers have lots of options in situations like that. You can even just turn the car off and hope you haven't picked up a vacuum leak.
I hear that these cars are "too smart" to let you shift while driving. I drive a stick (my preference) so I have no idea if that's accurate or not, but I did sit in a Prius at an auto show once and there's nothing mechanical about the drive/park/reverse selector. It merely indicates to the computer what you would like to do.
this is my sig
Correct for CAN. Though as for any protocol - garbage in; garbage out. The value(s) transported in CAN's payload may have been corrupted in memory or even in the CAN driver hardware* Actually, the "cosmic magic" is more likely to corrupt bits in static locations than represented as a voltage potential traveling along a differential bus with an active low being the dominant state. Of course, I have nothing to base this on, but I am posting here - *shrug*. * I designed both hardware and software for redundant CAN implementations.