Exponent: Gilbert's work not realistic - would make any car accelerate

Bloomberg report Toyota USA will present a demonstration (also available over the internet) that Gilbert's finding is wrong.

This is scheduled to take place on 8 March at Toyota’s U.S. sales headquarters in Torrance, California.

 

gilbert is a quack

 

Yes, it would be wrong to take Gilbert's work as 'this is what is happening.' However, Gilbert did do the 'forbidden experiment.' It is possible, not necessarily probable, that another experiment, say a resistive jumper from 12 VDC to Vcc might shift all sensor inputs up and lead to a similar result.

I see Gilbert in the same pattern as Prof. Langley versus the Wright brothers. Not all experiments succeed and we often learn much from the failures. Performing the experiment has a merit of its own.

Bob Wilson

 

Oh no! . . . not that!



.

 

Concerning the "merit" of Gilberts experiment?

Unfortunately, he does clearly have a right to execute the experiment, but conversely if the results of that experiment are going to be used in trying to implement Toyota further in accusation or speculation of a faulty or substandard electronic throttle system, then Toyota not only has right, but a responsibilty to respond.

It's my personal opinion that from what I've read, seen and "think" I understand, that Gilberts experiment is invalid as a "reason" to conclude that Toyota has a flaw in the electronics of their throttle system. This is primarily because I agree that he seems to be applying physical criteria and modification to the system and has not demonstrated how these changes could randomly or spontaneously manifest. If you are going to make accusations based on result of a very specific test, then it would seem to me that defining that aspect would be key.

But whether you agree with me, (and I could be wrong) or Gilberts results, the further problem is how it is being presented. ABC news, is not a proper venue for release of a results of a specific test, incomplete in definition and simply used as media tool to further damage Toyota's reputation and cause fear.

I'd have more respect for Gilbert if he had performed the tests, gathered results and avoided the media. IMO all he has done is show that there is a way to get The Toyota system to accelerate, but he has not demonstrated "how" these events could spontaneously self create in the real world. Taking an ABC reporter for a ride in a modified Toyota, flipping switches and "scaring" the reporter with "sudden acceleration" reduces his "experiment" to simple theatrics.

If you are going to make accusations based on results, I think you have a responsibilty to make sure those results are fully conclusive before making those accusations and that would include being able to demonstrate what conditions could, or would manifest in normal real world useage that could create the conditions that you are so eager to replicate for Television Camera's and the media's hungry eye. Otherwise you stray from the world of scientific experimentation into the world of David Copperfield and stage theatrics.

 

I agree. Gilbert's demonstration shows that an electronic system can be compromised, but his method is equivalent to fastening a string to a mechanical throttle linkage and pulling on it. Sure the engine accelerates, but is it a real world problem?

The good thing is that his demonstration makes people think about the reliability of systems. The bad thing is that the media takes this and feeds it to the uneducated masses, making them think that this is a real problem, instead of a simple demonstration.

Tom

 

Well, allow me to be a bit of a contrarian here:
The Toyota statement admits that Gilbert was able to compromise the system. Based on the news report, I can't see the evidence that proves it can't happen in the real world. They say it is highly unlikely--but remember, even if every report of unintended accelleration is absolutely true, that's probably less than one in a billion accellerations in a Toyota.
Now when I think about that thread on the Gilbert report where bob wilson and other looked at the thing, there was some concern that there was only 20 millivolts or so difference between two certain voltages, that could result in an error arising more easily than if there were higher differences between the voltages. I don't really understand that thing, but it made sense intuitively, and made me wonder if the fault is more likely to show up in a car where there was that small margin of error than in a car with a more robust system.
So, for all you out there who know a lot more engineering than I do (i.e., most folks here)--does it remain possible that the Gilbert error COULD pop up in most modern cars with electronic control systems, but that the Toyotas COULD be more vulnerable to it?
I'm very glad to hear at the end of the article that Gilbert and Toyota are continuing to talk, because maybe some good evidence will come out of it, and maybe Toyota will find a way to improve the product.

 

Someone posted the Exponent report "Evaluation of the Gilbert Demonstration" and it is an excellent analysis. In short, the probability is very, very small and there would be evidence in the vehicles. Yet no one has fund such evidence for this very improbable event. The difference between "possible" and "probable" makes life possible. For example, the air we breath in our respective rooms is subject to random, Brownian motion. It is possible that all air molecules could suddenly, randomly move to the corner of the room and we would asphyxiate in the vacuum. But this is improbable so we each can take another breath.

Bob Wilson

 

I have an honest question: I was under the impression that a GM car's ECU would throw an error code when the "Gilbert Demonstration" was done with its accelerator (implying that the error condition had been at some level accounted for by the egnineers who designed the car).

Why would we not expect the ECU in a Toyota to have a safe failure mode (error code) rather than a fairly dangerous one (acceleration), when supposedly-synchronized signals disagree, regardless of how rare the possibility might be? Am I mistaken about the GM part?

A previous thread had one poster ranting over and over about how rare the cases of unintended acceleration are - I distinctly remember that the poster went to a lot of effort to emphasize the word DENOMINATOR!!!1!one and the chances were something like 1 in 40,000 of it ever happening to you. Given Toyota's sales volumes, that makes 2 passenger cars a month, not including what you count in light trucks (SUV sales). At these large volumes, shouldn't there be an underlying expectation that very rare events occur on a somewhat regular basis in the engineering?

 

I don't know if anyone has tried.

Bob Wilson

 

Video: Expert recreates Toyota sudden acceleration

So.....? What's the deal?

 

The deal is for Gilbert, who was paid by Kane, who in turn was bought by a group of lawyers in the process of suing Toyota, to provide theatrical performance on ABC and in congressional hearing.

Now, I am not saying Gilbert is a liar/cheat/con-man. In fact, I believe Gilbert is a sincere and honest man at heart whose limited professional credential, seems to be miss used.

 

OK, so how does this change how one model ECU throws an error, and another just accelerates the car? I won't argue that it's kind of shady, but....safe failure vs not-so safe failure?

 

http://priuschat.com/forums/prius-h...unk-gilberts-throttle-claims.html#post1079365

 

That sir, I believe is best address by the auto industry and government at large.

So far, we have heard counter arguments with demonstrations from Toyota and its contracted investigator and academic associates, but no challenges from any other established entity, yet.

Also, we have not heard from NTSHA about their progress in investigating Mrs Smith's Lexus ES 350, which, was sold to (unsuspecting?) current owner. I am not sure if NTSHA had seek feedback from that owner, besides that of Mrs Smith herself ever since.

An lately, we read about "secretive" black box that only a prototype reader can decrypt. So here we are, another shady case based on shady assumptions.

Whatever happen to the committee's position on the subpoenaed documents from that ex-Toyota lawyer? The politicians were quick in drawing their guns during the hearing, but surprising slow in their investigation when documents are in their hands.

And that leads me back to the widely held suspicion: Is all these a witch-hunt?

 

Sure, why not? A 20mV differential is minimal and could be the product of any number of stray currents or resistive variations at any point in a complex system. HOWEVER, I personally do not feel there is an electro-mechanical issue, although no system should move into an uncontrolled state under such minor electrical variations in a life-or-death mission critical system. The reason why medical devices cost so much is that they are subject to insane quality controls and design qualification to validate their veracity.

There is also a lot of talk about errant high-energy particles or EMF interference contributing to possible problems. Most seem to forget the car itself is a Faraday Cage and that as such errant signals will be generally eliminated. (This is why antennas are on the outside and why a car can be hit by lightning and the occupants come out unscathed.) So if a cosmic ray hits a Toyota, it's unlikely to cause a problem.

 

If this were true, we would have no trouble with distracted drivers using handheld cell phones inside cars. Only cells with externally mounted antennas would work.

Cars have windows and other gaps. Windows in Faraday cages must be much smaller than the wavelengths that need to be excluded. The ratio of allowed window size to wavelength depends on the amount of attenuation needed. The two main cell phone bands have wavelengths of about a foot, the other bands and WiFi signals are much shorter.

Cosmic ray wavelengths are short compared to the spaces between atoms, so they go through 'solid' materials until they strike an atom's nucleus.