Regen causing unintended acceleration, according to new study?

[Three3]

This new scientific article, below, seems like a huge big deal. Basically it sounds as though EVs, including Teslas, may be interacting with Bosch brakes to in essence short circuit (especially when the brake lights error code is triggered) and accelerate due to loss of regen/addition of motor torque. Green has a whole thread going on at his Twitter about this too.

https://www.autosafety.org/wp-content/uploads/2015/03/Tesla-Regen-Brakes-and-Sudden-Acceleration.pdf

My question is, to those who perhaps understand the article better than I do, does this mean that we should all be turning regen to “low” until this is fixed (possibly via recall, or more hopefully, a firmware update)?

 

[mackgoo]

Didn't read the article. I've been driving Tesla's for 6 years and never had an issue.

 

[strykeroz]

The article is very detailed. However the missing piece is the hypothesis is not tested by taking a vehicle and simulating a brake switch failure combined with this regen response in controlled circumstances.

I couldn't actually see (but it might be there) confirmation the vehicle that hit the wall had a faulty brake light switch.

does this mean that we should all be turning regen to “low”

I don't believe that is explored either.

 

[brkaus]

I believe it’s BS generated to create doubt for a lawsuit/ settlement.

Looks like lots of quotes lifted out of context from here too.

of course, just my uneducated opinion.

 

[EnrgyNDpndnce]

Yeah I’m not buying this. Read part of the article but it’s so long I just didn’t want to invest the time. There’s a really good write up I’ve read before (wish I could remember source and link but I can’t) that details the redundant systems working in every Tesla to prevent unintended acceleration. The conclusion is, basically, it is virtually impossible for unintended acceleration to occur in a Tesla. Accidental acceleration can and does occur, when drivers THINK they are pressing the brake pedal but are actually pressing the go pedal. This has happened for years in many makes and models.

 

[LACALawyer]

The "unintended" part of "unintended acceleration" means you didn't intent to mash the go pedal when you were aiming for the brake.

 

[arnolddeleon]

The article is very detailed. However the missing piece is the hypothesis is not tested by taking a vehicle and simulating a brake switch failure combined with this regen response in controlled circumstances.

I couldn't actually see (but it might be there) confirmation the vehicle that hit the wall had a faulty brake light switch.

I don't believe that is explored either.

Yeah, with such a simple theory (brake switch failure) it should be easy to build a test scenario and attempt to reproduce the problem.

For those don't I want to slog through the thing here is the summary and conclusion:

The EDR log data from a Tesla sudden acceleration incident has been analyzed. To explain the EDR data, the operation of Tesla’s drive motor control system and braking system were examined. As expected, friction braking and regen operation are completely separate with no blending, with the possible exception of Tesla models having a single rear drive motor. The braking system, however, also includes several vehicle stability control functions that can have a profound effect on regen operation in the presence of wheel slip, such as stopping regen when going over bumps and while turning corners. One of these slip control functions called Electronic Drag Control (EDC/MSR) can even cause the drive motor to speed up if regen is causing slip in the rear drive wheels that can lead to an oversteer or understeer. This same slip control function can be misled by a defective brake light switch to confuse a brake-induced deceleration for a regen-induced deceleration, in which case as the driver presses harder on the brake pedal, a larger positive motor torque is produced. This is believed to be the cause of sudden acceleration in the incident under consideration. To see if this same mechanism could explain other Tesla sudden acceleration incidents, NHTSA contact reports were examined. It was found that over 70% of all Tesla sudden acceleration incidents could be explained by this cause. Similarities between the braking systems for Tesla and non-Tesla vehicles imply further that this same cause can explain many sudden acceleration incidents in non-Tesla electric vehicles having rear-wheel drive (RWD) or all-wheel drive (AWD). Finally, it is concluded that vehicles having internal combustion engines along with rear wheel drive are susceptible to sudden acceleration from this cause if the brake lights are defective. Comments for testing vehicles for this mechanism are provided​

 

[Vines]

If you dig into the article it makes some very good points.

Certainly some hardware or software on the Model 3 failed in this case, assuming you believe the data.

Couple unanswered questions are things like whether the Model 3 has a pair of redundant physical brake light switches and whether it explicitly checks both of them through the whole ABS cycle. Now I want to pull one of these switches apart myself.

Also, the government estimates 16000 times per year a SUA crash happens from a driver accidentally hitting the wrong pedal.

if 400,000 Model 3 are on the road, then that is about 0.15% of the registered passenger vehicles on the road. Tesla Model 3 are expected to have 0.15% of SUA, that are the drivers fault per year, which the government estimates at 16k per year. Approximately 24 SUA events happen in a model 3 per year.

In the article, 102 events were recorded over a 7 year period. Statistically 168 events would have happened due to driver error. I am skeptical that this was not discussed at all. Even after the crash, most SUA drivers report that they were certainly hitting the brake pedal, even when the data and video shows otherwise.

It should be simplicity itself to test this scenario, and if investigation is really the point why didnt the author address that or perhaps test that.
Instead some of his evidence is TMC forum posts lol.

 

[Three3]

should be simplicity itself to test this scenario, and if investigation is really the point why didnt the author address that or perhaps test that.
Instead some of his evidence is TMC forum posts lol.

From what I understand, this is a retired (in his late 70s I believe) engineer who volunteers his time with this nonprofit and sort of does these papers as a hobby. I don’t think the onus is (or should be) on him or the organization to do this sort of testing, since he’s already providing a valuable service here.

I initially believed Tesla’s position but this study at least deserves some attention from regulators, Bosch (since seems mostly to be an issue with their brakes) and Tesla. Even if you just read the appendix, there is a shocking degree of similarity between many of these incidents: Usually someone making a sharp low speed turn into a parking spot and away she goes.

What seems to be happening isn’t that Tesla or the brakes are malfunctioning per se but that the communication between the brake and regen model isn’t there.

Ie, if I understand the paper correctly, that there are instances in which ABS is maybe coasting and the car reads this as a wheel slip and so lessens regen, aggressively so at times, which may lead to actual acceleration, which then may cause the ABS to further let go.

Important to note that this appears to be an issue with all EVs that use these brakes, not just Tesla. (I am not trying to stir up any FUD here, just hoping for a speedy fix or at least some rigorous testing proving the paper’s conclusions untrue. My hunch is that in the short term at least, the software could be fixed so that a brake light fault would make the vehicle inoperable, which would appear to prevent ~70% of such incidents.)

 

[MP3Mike]

There is someone who works in the industry that says this study is completely wrong and he is willing to bet a million dollars on it: https://twitter.com/BrownSoylent/status/1273931484676665351

 

[SlimJim]

I don't use brakes much.

 

[roblab]

My take? Someone mashes the accelerator instead of the brake. Then they his something. Now they want to sue.

These suits seem to always fail.

 

[1.21GW]

It’s funny that vehicles have unintended acceleration Pulling into parking spaces. Typically just as people are trying to pull into a parking spot (and the same time they are supposed to hit the brake), the vehicle takes off! I mean, what are the chances!!??

 

[Vines]

I wonder if some of the issue is with 1 pedal driving, people subconsciously associate the accelerator with the brake pedal.

I don't know, but reading the sensor data in the article paints a strange picture. Something isn't stirring the kool-aid.

 

[MP3Mike]

From what I understand, this is a retired (in his late 70s I believe) engineer who volunteers his time with this nonprofit and sort of does these papers as a hobby. I don’t think the onus is (or should be) on him or the organization to do this sort of testing, since he’s already providing a valuable service here.

So you don't think someone making such significant allegations should bother to test their theory before releasing the information publicly? (Look at all of the bad "scientific" studies related to COVID-19 that have had to be pulled because they were bad and contained false conclusions.) If his conclusions are false then he isn't "providing a valuable service here." (Well maybe he is to the FUDsters and the people suing Tesla that have wrecked their cars.)

I could see that if the testing was involved and expensive that you might give them a pass, but if testing his theory is really a matter of disconnecting the brake switch and driving in a parking lot for a few minutes it seems irresponsible to not have done that before releasing the paper. And sure maybe since he is volunteering, if true, he shouldn't have to do that, but the non-profit should have.

Even if you just read the appendix, there is a shocking degree of similarity between many of these incidents: Usually someone making a sharp low speed turn into a parking spot and away she goes.

Yep, many people confuse the pedals. Especially when they are slowing down by regen with their foot over the accelerator and think I need to slow down more, so I need to push on the pedal... But the wrong one.

(I am not trying to stir up any FUD here, just hoping for a speedy fix or at least some rigorous testing proving the paper’s conclusions untrue.

Why is it Tesla's responsibility to spend bunches of money disproving every crack pots theory? (That they haven't even tried to prove themselves.)

My hunch is that in the short term at least, the software could be fixed so that a brake light fault would make the vehicle inoperable

Have they even bothered to check if any of these cars have a failed brake switch? It sound more like they are just assuming it does because the person said they pushed the brake pedal but the computer didn't register it.

 

[Three3]

Rather than finding fault with the scope of this particular paper (and he talks about the complexities of testing the brake fault hypothesis), I believe the safe and responsible thing would be for Tesla and Bosch and the other companies (let alone the regulators) to at least look at and consider his conclusions and test them.

The guy clearly knows what he’s talking about, and his hypothesis seems at least worth further study and testing. Science works by others testing and retesting and validating (or invalidating) original research. No original study is ever the final say.

In summary, the guy has a reasonable hypothesis that I would like to see proven or disproven, regardless of who does that future work. The worst case of testing said hypothesis (or at least refuting, line by line by line, with solid evidence) is that it’s a waste of time and money. Worst case of NOT carefully examining said hypothesis is that a major vehicle safety flaw goes undetected.

I’ll take my chances with the former.

 

[M109Rider]

This guy is retired and decided to delve into the world of unintended acceleration because he believed Toyota had a problem. turned out he was wrong, but kept pursuing his theories on unintended acceleration anyway.
In my opinion, this is highly educated guy, that enjoys spending his tine in this area. No issues with that.
But I don’t believe there are any issues to find. At least not with Tesla.

Here is a copy and paste from questions he was asked about this whole hobby of his into unintended acceleration.

“Question: Dr. Belt, if your first two papers were wrong about the cause of sudden acceleration, is it possible that your latest two papers are also wrong about the cause of sudden acceleration?
Answer: Yes, it is possible that my latest two papers are also wrong because the theory has not been tested at this time. However, it is difficult to come up with any plausible theory of sudden acceleration, so when is one is able to come up with a theory that explains so many of the observations associated with sudden acceleration as this one does, one feels that the theory must be largely correct. There may be slight changes needed in how the negative voltage spike produces an incorrect voltage compensation value, like maybe the voltage spike causes a memory upset instead of just being sampled incorrectly, or whether there is only one or possibly two compensation values. But overall, I believe that the theory is better than 90% accurate. I was unable to test the theory myself because of the expense involved with getting a suitable vehicle and with getting the test equipment needed. But I have provided detailed instructions for testing the theory in the most expeditious manner possible. I hope that someone is able to do this as soon as possible.”

 

[Three3]

Honestly this guy sounds like a pretty good scientist to me. He has come up with a theory, acknowledged that he may well be wrong, and is asking others to test his theory (which is proper; even if he did test his theories himself, someone else should retest).

 

[MP3Mike]

Honestly this guy sounds like a pretty good scientist to me.

You call being 100% wrong on your two previous theories being a pretty good scientist?

 

[Three3]

You call being 100% wrong on your two previous theories being a pretty good scientist?

Yup. I think you misunderstand how science works. It’s not like on the movies. It’s tough work and more disproven hypotheses than proven ones. Fits and starts and dead ends abound.

No one has convinced me (or will) why his conclusions in this article do not at least deserve careful examination, given what’s at stake. To flip this on its head, if you are so sure he is wrong, I think you’d look forward to having it be disproven. But in any case, and luckily, folks on Internet forums (myself included) do not set public policy, and I’m just grateful for this guy and his work and I assume it will at least get noticed by regulators and manufacturers, who will indeed consider his claims carefully.

My original question was IF his hypothesis were accurate, would we be safer to keep Regen in low or standard. I think everyone who has done a deep dive into the article is correct in saying that there is no way to glean that information from this article (and it doesn’t seem to be a question he was looking to answer anyway).