Tesla: Model X Accident Caused By Driver Error, Not Autopilot (computerworld.com)

An anonymous reader writes: Tesla has responded to a recent report from a Model X owner claiming their vehicle suddenly accelerated at "maximum speed" by itself, jumped a curb and slammed into the side of a building while his wife was sitting behind the wheel. They said it analyzed vehicle logs, "which confirm that this Model X was operating correctly under manual control and was never in Autopilot or cruise control at the time of the incident or in the minutes before. Data shows that the vehicle was traveling at 6 mph when the accelerator pedal was abruptly increased to 100%. Consistent with the driver's action, the vehicle applied torque and accelerated as instructed. Safety is the top priority at Tesla and we engineer and build our cars with this foremost in mind. We are pleased that the driver is ok and ask our customers to exercise safe behavior when using our vehicles." When will people stop lying about Tesla's Autopilot mode crashing their cars? One Tesla owner recently filed a Lemon Law claim against the company over a high number of quality control issues.

Comments

No one hurt .

[invictusvoyd]

slammed into the side of a building while his wife was sitting behind the wheel.

hmm ok . Happens.

Re:No one hurt .

[michelcolman]

You may laugh, but, from the actual article:

"researchers found that there were seven to 15 crashes per month in the U.S. caused by pedal application errors. Females were the drivers in nearly two-thirds of the pedal misapplication crashes identified in crash databases and in a media scan used in the study."

Re:No one hurt .

[prefect42]

This isn't something I can imagine doing anywhere near as badly in a manual. You panic, you stomp brake and clutch. Miss the brake and go for the accelerator, and you rev like crap but don't accelerate. You miss the clutch, you stall it. Seems like quite a challenge to miss the clutch and hit the foot rest, whilst simultaneously missing the brake and hitting the accelerator.

http://www.nhtsa.gov/staticfil...

Researchers reviewed each crash narrative to determine whether the crash actually resulted
from a pedal application error. Of the 2,930 crashes, 2,411 were caused by a driver applying the
accelerator when he or she intended to apply the brake. Fifty-eight were the result of the driver’s
foot slipping from the brake and pressing the accelerator, 47 were the result of the driver pressing
the wrong pedal in a vehicle with manual transmission (either clutch or accelerator rather than the
brake, or the brake rather than the clutch). Reviewers determined the remaining 414 crashes not to
11 be the resultt of a pedal misapplication; these 519 incidents were therefore excluded from the present
analyses.

Re:No one hurt .

[michelcolman]

Keep in mind that the US has very, very few cars with manual transmissions. So 47 is probably a pretty large percentage compared to the other numbers.

Re:No one hurt .

[houghi]

It is, I think, just how humans function as a human under stress. They want to slow down and in an automatic that means pushing down on a pedal. If you have the wrong pedal, you will accelerate. That means your brain says :push harder on that pedal to slow down and you go even faster till you either hit something or you realize you are pushing the wrong pedal.

In a shift car, slowing down means taking your foot OFF the pedal and shift down and back on the pedal. That means you will correct the handing of the wong pedal. Also as shifting alread mean taking the foot off the accelerator, it means you will go a little bit slower, not a liittle bit faster.

Note that I am talking about slowing down, not about breaking. In short: In an automatic you push down to go slower, in a shit car, you pull up and then push down.

OK. Enough cars, a computer comparison.
The difference in typing 'rm -rf /'ENTER or 'rm / -rf'ENTER. In the second you have just a bit more time to realize you type / and not /tmp. That time is the second it take to type "-rf". In the first you have hit ENTER and it is too late.
Same happens with the car. In a shift car you have just a bit more time to realize what is going on and correct that behaviour.

Re:No one hurt .

[necro81]

I've only ever owned manual transmission cars. I've always liked the ability to feather the engine or disconnect it from the wheels entirely. Even today, the clutch and manual transmission are almost always mechanical assemblies, not fly-by-wire. When Toyota was having all those issues with unintended acceleration, I'll admit that I felt smug, knowing that I had the ability to disconnect the engine, coupled with decades of experience that makes depressing the clutch instinctual.

For various reasons, my next car is likely to be a plug-in hybrid or pure electric. I'm going to miss that capability.

Re:No one hurt .

[beelsebob]

Given that no current EV on the market has a gearbox at all, that seems pretty tricky :P

Re:No one hurt .

[Anonymous Coward]

Really? Every EV has a 2:1 reduction gearbox on the motor...

Re:No one hurt .

[galgon]

The Model X has auto-pilot but not collision detection? Shouldn't the car have seen the wall and applied the brakes no matter how hard she stomped on the gas?

Re:No one hurt .

[eriklou]

You may laugh, but, from the actual article:

"researchers found that there were seven to 15 crashes per month in the U.S. caused by pedal application errors. Females were the drivers in nearly two-thirds of the pedal misapplication crashes identified in crash databases and in a media scan used in the study."

And how many of those women were wearing something other than a standard shoe...

Re:No one hurt .

[beanpoppa]

I drive manuals, too. But they do have their problems in this type of situation. I had a friend who was rear-ended by a manual driver, twice. A driver first hit the clutch, but missed the brake and rolled (at a rather high speed) into him. Then, after the hit and rebound, her foot came off the clutch, at which point the engage engaged again, and lurched her back into him.

Re:No one hurt .

[beelsebob]

While you're technically correct (the best kind of correct) the intention was obvious - no shipping EV has a gearbox that actually changes gear ratio while you're driving.

Re:No one hurt .

[Smilodon]

Actually, there is at least one EV shipping with a gearbox. Clutch and all!

http://www.victorymotorcycles....

And it's a pretty cool one too!

Re:No one hurt .

[Grishnakh]

In short: In an automatic you push down to go slower, in a shit car, you pull up and then push down.

You think cars with manual transmissions are shitty? Why did you need to add this kind of vulgar, biased language to what was otherwise a pretty good fact-based argument?

Re:No one hurt .

[Dcnjoe60]

User control overrides almost all car controlled actions. If she would have removed her foot off of the accelerator, the car would have likely seen the wall and applied the brakes. At least that's how it works in all the current model cars with collision detection.

Hopefully, you are incorrect. If so, then collision avoidance is almost useless, at least in traffic as it is likely that a distracted driver would not see the obstacle or other vehicle in time to remove one's foot. The cars I have driven with this feature will brake, quite aggressivley, I might add, even if one's foot is on the accelerator.

Re:No one hurt .

[GLMDesigns]

I think he meant to write "shift."

See previous sentences "In a shift car, ..."

Re:No one hurt .

[AaronW]

Usually it's more than 2:1. On my Tesla model S P85 it's 9.73:1 since electric motors typically have no problem spinning at a much higher speed than an internal combustion engine.

Re:No one hurt .

[stooo]

Drive Manual cars.
Use Linux.
Just live.

Re:No one hurt .

[R3d M3rcury]

You panic, you stomp brake and clutch.

When I was 15 and learning to drive, we had driver's ed classes in school (on an automatic) and my parents were also teaching me (on a manual). So I was going back and forth.

My favorite clumsy moment was when I was pressing the brake to stop at an intersection and the car started sputtering, so I immediately took my right foot off the brake and pressed down the clutch. Now we're rolling towards the intersection and my Mom is screaming for me to stop so I take my left foot, cross it over, and press the brake.

So here I am, sitting at the intersection, with my legs crossed and I'm thinking, "How do I get out of this mess?" My right foot is on the clutch and my left foot is on the brake. If I release the clutch with my right leg, in order to get my feet untangled, we stall. If I release the brake with my left leg, the clutch pedal is too small for me to easily transfer it and free my right leg. Meanwhile, my Mom has figured out my dilemma and is laughing her ass off. She finally recovers enough to suggest I put the car in neutral so I can release the clutch and get my feet all sorted out.

So, when you panic and press the clutch and the brake, make sure you do them with the proper legs.

Re:No one hurt .

[michelcolman]

So what happens if you are overtaking a car on a two lane road, another car is coming towards you from the other direction but you judge that you can still make it with room to spare, so you add more gas, and meanwhile your car sees the other car coming towards you, has no idea that you are going to move out of the way, and therefore applies the brakes?

Manufacturers prefer (and should prefer) having a few false negatives where the car does not brake to avoid a collision (it's still your fault after all), rather than have false positives where the car's braking actually causes an accident.

The main purpose is to avoid crashes due to the slow reactions or distractedness of the driver. If the driver suddenly stomps on the gas, that's a different situation and the car should obey. The car should not assume it's smarter than the driver (even though sometimes it might be).

Re:No one hurt .

[Dcnjoe60]

So what happens if you are overtaking a car on a two lane road, another car is coming towards you from the other direction but you judge that you can still make it with room to spare, so you add more gas, and meanwhile your car sees the other car coming towards you, has no idea that you are going to move out of the way, and therefore applies the brakes?

Manufacturers prefer (and should prefer) having a few false negatives where the car does not brake to avoid a collision (it's still your fault after all), rather than have false positives where the car's braking actually causes an accident.

The main purpose is to avoid crashes due to the slow reactions or distractedness of the driver. If the driver suddenly stomps on the gas, that's a different situation and the car should obey. The car should not assume it's smarter than the driver (even though sometimes it might be).

Hopefully the AI in the car is able to calculate the speed of the on-coming vehicle and your speed quicker than you can and if it is safe to take the course of action you describe it will allow it, but if not, it will prevent it.

Re:No one hurt .

[michelcolman]

It's not quite there yet. Not by a long shot. But some day, sure, it will.

Re:No one hurt .

[jimtheowl]

It is not Slashdot, but an Oklahoma jury that determined that the automaker acted with “reckless disregard” regarding their firmware. We also have an admission of guilt from Toyota regarding that issue.

And by European cars, I presume that you are referring to the BMW fiasco.

Would you suggest that we blame all equally regardless of guilt?

Re:No one hurt .

[ChrisMaple]

Disengaging the clutch while slowing down depends entirely on circumstances. It's not the normal operation when slowing from 70 mph to 50 mph, and it's entirely superfluous in a panic stop.

Re:No one hurt .

[rpstrong]

And the sound of an automatic is, "Whooosh".

Re: No one hurt .

[necro81]

Ooooooh. Anonymous Coward calls BS on Slashdot without offering anything resembling counterargument or evidence. That sure puts us in our place!

Really?

[ledow]

Not being funny...

but if the logs show 100% acceleration, that just reflects the sensor value. Not that the user - or indeed anything else like a dropped handbag - actually pressed the pedal that far.

Although I'm always the one to shout "user error" first, and that's quite likely in this case, the logs alone are not sufficient to prove fault. Only to act like a flight recorder and say what the sensors recorded and what the machine did in response to that input.

How the sensor got that reading could still be manufacturing fault, cable fatigue, or a million and one other things not the fault of the driver.

Re:Really?

[invictusvoyd]

How the sensor got that reading could still be manufacturing fault, cable fatigue, or a million and one other things not the fault of the driver.

Your argument is valid but "his 45 year old wife behind the wheel" gets priority .

Re:Really?

[E-Rock]

I think the bigger point is that the car wasn't in autopilot mode at the time. I don't think the drivers are realizing that they can check and call them on their bullshit.

Re:Really?

[willy_me]

How the sensor got that reading could still be manufacturing fault, cable fatigue, or a million and one other things not the fault of the driver.

Designing a pedal sensor that errors to 0% is expected. So when one of those million things goes wrong you do not get the 100% acceleration experienced in this situation. A far more likely scenario is that something dropped onto the acceleration petal. Alternatively, when in a state of shock, the driver mistook the acceleration petal for the brake.

Re: Really?

[thesupraman]

She claimed it was in autopilot.. It appears it was not. How does a sensor reading explain that?
It's called arse covering and blame shifting.. People do it all the time.
If she had described an unexpected acceleration while manually driving then the story may wash..

Re:Really?

[Balthisar]

In the spirit of advocacy for the Devil, one of the issues that I had with my 1985 Ford Escort was the throttle position sensor. Luckily applying 100% torque to a 1.6L, 70 HP engine only produces 119 Nm of torque (vs. over 900 Nm for the Tesla). Hmmm... weight, about 1000 kg vs about 2200 kg, still a pretty big difference. I'd rather have the bad TPS in a 1985 Escort than at 2016 Model S.

Certainly in 1985 the TPS position probably wasn't logged. Mass market fuel injection for cheap cars was still in its infancy, and the signal probably went directly to the throttle body, maybe via detour to the ECM (if there was one). The Tesla, though, is obviously logging, but I wonder at what frequency? If the frequency is fast enough, then we can observe the curve of the TPS as the driver engaged it, i.e., how fast was it pressed. If it's just a flaky TPS, then we might see an instantaneous change. If the sampling frequency is too low, though, we won't be able to distinguish them.

Re: Really?

[SuperDre]

Yep, and people always tell the thruth to what really happened........ NOT!

Re: Really?

[qbast]

She claimed it was in autopilot.. It appears it was not..

She claimed, Tesla claimed otherwise. Actually I would expect that in case of some strange error, the autopilot disengages immediately so Tesla can truthfully says it was not on during time of accident.

Re:Really?

[Anonymous Coward]

If my car suddenly accelerated, I would call that "on autopilot" in a car that has an autopilot. Users are not engineers, especially not 45 year old wives. By the way, sudden acceleration is not an unusual "cause" of car crashes. It is typically caused by (old) people mixing up the gas and brake pedals, and of course, once the car starts accelerating, they push the (wrong) pedal all the way down. Afterwards they report braking as hard as they could and the car accelerating nevertheless, i.e. they insinuate technical fault, even in low-tech cars. It's not lying. They report it as they experienced it.

Re:Really?

[Viol8]

Indeed. To me this has all the hallmarks of insurance fraud and/or an attempt to get off negligent driving charges.

Re:Really?

[slashrio]

...but if the logs show 100% acceleration, that just reflects the sensor value.

Well, then there is also the damage to the building which confirms that there must have been some speed, hence accelleration.

Re:Really?

[Mashiki]

A lot of cars still don't log values at the TPS, it'll report with a tool connected but that's it. A buddy of mine owns a garage and recently had a car(2013 enjoying those 3 year warranties up here in Canada yet?) in and he couldn't figure out what the problem was, at times the car would stall at others it would would run at the maximum RPM under no motion(around 4200RPM) according to the vehicle speed sensor and sometimes it would be just fine for days on end. TPS sensor showed no errors to the ECM either. I mentioned he should manually check the resistance values and surprise...bad sensor, zero logged codes. And of course the flow charts don't recommend manually checking the values because the ECM is supposed to report whether or not it's bad.

Not the first time I've seen that either, my parents had a pontiac transport 3.1L which was made during the throttle body injection/MPFI cross-over with other engines(3.8L were MPFI for example). Same problem with the TPS on their van. I've even seen the TPS throw errors because of the coolant temperature sensor too(different then the actual inside temperature gauge sensor). Saturns made in '93-01 were famous for that, it was a $23 part but would cause TPS errors at the ECM side, but resistance values were good. So yeah two sensor, the one mounted in the intake manifold would report fine(it also reported values to the gauge), but the engine senor mounted in the engine block would report bad values.

Re: Really?

[Anonymous Brave Guy]

If you're building a safety critical system, having two sensors won't do you much good. Even if they disagree, you have no way to know which is wrong, only that some fault exists. The best you can do in that situation is try to fail to safe, but in a vehicle where stopping suddenly at the wrong time or in the wrong place could be more dangerous than carrying on until it's safer, there is no perfect failsafe mode.

Having three parallel systems, ideally made with different types components to guard against design defects, gets you something more useful, since you can at least take a majority vote if one of the sensors is out of sync with the other two. Obviously this is also more expensive to implement, though.

Re:Really?

[Anonymous Brave Guy]

It might not have been in whatever they officially call autopilot, but modern cars suddenly accelerating due to control system failure is, unfortunately, not unheard of.

Of course, neither is someone confusing the accelerator and brake pedals while in a panic that the car is suddenly speeding up. We don't know enough to say what really happened here from what I've seen so far.

Re: Really?

[Opportunist]

That's exactly the problem. What you suggest is very ok for a person-graded orbital system or maybe a passenger plane, but who's going to pay for it in a consumer car?

Re:Really?

[Opportunist]

There's nothing preventing "45 year old wives" from knowing something about the machines they operate. Actually, I do expect people operating machinery that can lead to accidents that endanger the lives of others to know enough about them to operate them safely. Independent of gender, race, age... you get the idea.

Re:Really?

[DrXym]

Depends what Tesla thinks by the term and what some driver might think if their car suddenly developed a mind of its own and started uncontrolled acceleration. I might call it autopilot too (in a Death Proof kind of way) if it happened to me. Not saying that's what happened but its certainly possible.

Re:Really?

[ADRA]

Rapid acceleration due to controller failure AND the lack record for brakes being applied at the same time? That sounds unlikely. Possible, but unlikely.

Re: Really?

[Anonymous Coward]

Potentiometers are cheap

Re:Really?

[sociocapitalist]

I think the bigger point is that the car wasn't in autopilot mode at the time. I don't think the drivers are realizing that they can check and call them on their bullshit.

Generalizing a question, does a certain 'mode' have to be activated for a bug in software/firmware to cause a problem with a device that might currently be in another mode?

I would say no.

To give a legacy car analogy, cruise control can be disabled and the gas pedal could still get stuck at full acceleration.
http://abcnews.go.com/Blotter/...

Re:Really?

[Rei]

Indeed. Accelerators and brakes are very heavily regulated parts of a car, as they're safety critical systems, and each must be designed to have its default failure mode be in a "least likely to cause an accident" manner. With accelerators, that means a failure should always be interpreted as "no throttle".

Re: Really?

[Anonymous Coward]

But capitalists are cheaper. Look into the process of designing a modern car...

Re: Really?

[michelcolman]

Majority voting with three systems has turned out to be dangerous on quite a few occasions. At least with two systems disagreeing, you can decide "I don't know which one is right, so I'll do the safe thing". With two out of three giving an incorrect value, it leads to misplaced confidence in this wrong value.

Examples:

- (human): Primary altimeters different between captain and F/O, standby altimeter agrees with captain, but F/O altimeter was correct. Don't remember the flight number or even the company, but they crashed into a mountain. If they hadn't had the standby altimeter, they would have gone with the lowest of the two indications until they could intercept a glide slope somewhere.

- (automatic): The A320 has independent Angle Of Attack probes, but if you climb through heavy icing, it turns out two can freeze up at the same time. The system sees a confirmed stall (even though it should be impossible given the current speed, attitude and inertial measurements) because two systems agreeing can't ever be wrong, and it pushes the nose down. Even if both pilots pull back on the stick, the nose continues to drop. We now have an official procedure to shut off two of the Air Data Reference units, leaving only one remaining. That's the only way of shaking the absolute confidence of the system in two identical but false signals.

Re:Really?

[AmiMoJo]

It's likely she just assumed or said it was autopilot when the car accelerated without her commanding it to. Maybe she had the mat laying over the accelerator, maybe she pushed the wrong pedal, maybe the sensor failed.

This is the nature of bug reports from users, vague and sometimes misleading but not always just user error.

Re:Really?

[NotInHere]

Well if it indeed was sensor fault, then given a large enough fleet size it will happen more than once. And once it starts to happen multiple times, Tesla is in trouble, having them to recall everything etc.

Sensor fault can be easily solved though simply by placing a second sensor, that confirms the first sensor's readings. Fun starts of course if the sensors show opposite readings, so better have three and trust the majority, while showing the driver that there is a fault with one of the sensors.

Re: Really?

[Ichijo]

in a vehicle where stopping suddenly at the wrong time or in the wrong place could be more dangerous than carrying on until it's safer, there is no perfect failsafe mode.

In the absence of someone tailgating you, when is controlled braking dangerous?

Re: Really?

[Rei]

Three sensors is not necessary; two (plus fault logging) is fine.

Accelerator: chose the lower value of the two, log an error if they differ by more than a given margin.
Braking: chose the higher value of the two, log an error if they differ by more than a given margin.

The worst case for the former is that your car doesn't accelerate. The worst case for the latter is that your car slams on the brakes. In neither case would you be at fault in an accident even if there wasn't a hardware error, as the person behind you is always supposed to maintain sufficient distance to avoid an accident regardless of what the person in front of them does.

Also, I find it kind of strange how people treat drive-by-wire as if it's some inherently erroneous system, yet think driving by mechanical linkages is somehow foolproof.

Re:Really?

[Opportunist]

At the very least I expect someone to be able to operate a machine so that it is of no harm to me. If the person is unwilling or unable to learn enough about the machine to not be a threat to others, the person has no right to operate the machine.

Re:Really?

[Opportunist]

Most of all I don't use more words than absolutely, positively and ultimately required to express the message that I wish to convey.

So you are right. The word I should have used is "demand".

Re: Really?

[n0creativity]

Would you really wanna be in the middle lane of a 5-lane, 65 mph, highway and have your car 'controlled brake' to a stop? What about when your pulling out of a parking lot, turning left across a busy 2 lane road? Crossing railroad tracks? On an icy road? There are quite a few situations where I wouldn't want my car to begin a controlled braking action. I would tend to agree that there is no 'perfect' failsafe action that does not take into consideration the context of the situation.

Re: Really?

[Anonymous Coward]

Almost all gas pedals in cars are implemented as potentiometers. (Not only Tesla. This is being done for years now.)

A pot-meter has 3 legs and both resistor values are measured.
If the measurements don't agree, an error is logged in hte on-board computer (and warning is generated) and the safest value is assumed.

So in case of a 100% acceleration one measurement would be "low" and the other "high", the exact oposite of not touching the gas pedal.

If you are wondering: The AD convertors are also separated in hardware.

In other news: In electronically controlled cars, applying the gas pedal and the breaks at the same time, get you an error and only breaks...

Re:Really?

[K. S. Kyosuke]

If my car suddenly accelerated, I would call that "on autopilot" in a car that has an autopilot.

And if your car didn't have it, you'd have to find a different excuse.

Re: Really?

[Buchenskjoll]

Yes, three sensors. Like The Minority Report.

Re:Really?

[grahamtriggs]

Users are also capable of not telling the truth. The logs are almost certainly accurate.

That doesn't entirely rule out a fault - if the system erroneously reads a 100% accelerator pedal depression, then it will record that and act on it; the error then being in the sensor, not the logging or action taken by the car.

But when somebody is parking, they are going to press the brake to stop - and if they find they are not stopping, or accelerating, they'll press it harder. So a 100% depression is also consistent with someone mistakenly pressing the accelerator instead of the brake.

When you make it possible to the blame the car, some people are going to do so instead of taking responsibility. On balance, I'm inclined to believe this was user error.

Re: Really?

[geoskd]

Yes, three sensors. Like The Minority Report.

Yes, because anecdotes are how all real science and engineering is done.

Re:Really?

[tomhath]

FTFA:

suddenly and unexpectedly accelerated at high speed on its own

I don't see any mention of autopilot in TFA. Maybe he claimed it was on autopilot somewhere else, but not here.

That said, everything about this indicates she stepped on the accelerator instead of the brakes.

Re: Really?

[Anonymous Coward]

I'm not sure if the semi behind me smashed my car and me in it into small bits I'd be much comforted by it not being my fault

Re: Really?

[Jarik C-Bol]

The Raman's always did things in threes, and look how that turned out.

Re:Really?

[drinkypoo]

Data shows that the vehicle was traveling at 6 mph when the accelerator pedal was abruptly increased to 100%.

if the logs show 100% acceleration, that just reflects the sensor value. Not that the user - or indeed anything else like a dropped handbag - actually pressed the pedal that far.

You are assuming that they don't have sufficient scanning resolution (in the temporal domain, that is) to determine the difference between a bad spot in the potentiometer/problem with the optical sensor (whichever it is) or a wiring fault, and the user smashing the accelerator pedal to the floor. I don't know that they do, but you don't know that they don't. With proper logging, such a problem would be detectable from the log.

Re:Really?

[Applehu Akbar]

"...evaporate on their 44th birthday because God abhors an old, married, female engineer."

Actually, management requires that all engineers evaporate on this date.

Re:Really?

[drinkypoo]

Technology is about to be mandated that will stop incidents like this; mandatory Automated Emergency Braking is coming in what, 2020? 2025? Real soon now, since all the major manufacturers and suppliers have an AEB solution available. And then "wrong pedal" collisions will be a thing of the past. The car simply won't let you do that.

It's too bad enough of us couldn't be responsible for humans to be able to keep the responsibility of driving, but more and more of it is being taken away.

Re: Really?

[drinkypoo]

Note how these logs are only allowed to be analyzed by the people that have the most incentive to doctor them.

It's called a Subpoena, and it happens every day. You file the case, you get the logs, you find out if they are accurate.

Re:Really?

[Assmasher]

Sure, if that sensor reading went from 0 to 100% (or similar) and then stayed registering 100%. It seems very likely, given Tesla's response that there is a very 'human' curve in the application of the accelerator prior to the accident and then a very 'human' curve to the release of the accelerator post impact. Then, a surprisingly totally functioning and normal accelerator afterwards...

Re:Really?

[drinkypoo]

With accelerators, that means a failure should always be interpreted as "no throttle".

That only works if you can detect the failure. The sensor can fail in such a way that it looks like it's going to full simply because it is damaged. The obvious answer is to use two sensors, each with their own leads, and report a failure if just one of them goes bad. This is what my car (1997 A8Q) does with the brake switch; it's actually two switches in one. It doesn't have throttle-by-wire, though, so it only has one TPS.

Re: Really?

[dcw3]

The Raman's always did things in threes

That's really using your noodle.

Re: Really?

[drinkypoo]

I find it kind of strange how people treat drive-by-wire as if it's some inherently erroneous system, yet think driving by mechanical linkages is somehow foolproof.

Mechanical linkages can be grabbed and wiggled to be evaluated. Nothing will tell you if an electrical component is about to fail.

Re:Really?

[dcw3]

Your expectation does not match the legal requirements in the U.S. While desirable, too many people just don't give a damn. We should have fixed this situation long ago, but fortunately, we'll be saved in the not too distant future by self driving vehicles.

Re:Really?

[dcw3]

Having not RTFA, I'm wondering about this 100% reading, and how frequently log entries are made. You don't go from a full stop to 100% normally...the accelerator isn't an on/off switch. No, people don't depress the accelerator to the floor, especially going in reverse. And, even if you thought you had your foot on the brake instead, you wouldn't be attempting max brake pressure w/o some prior motion. It just doesn't seem to add up, in spite of my knee jerk reaction that it's likely driver error.

Re:Really?

[Woek]

I was thinking the samen thing. The transient of the pedal command would however be very informative. If the sensor value increased in a way consistent with someone pressing or slamming the accelerator, it would point to a user error. If the value increases to 100% within one sample time, that would point to a sofware or electronics error. At least, if the sample rate is high enough...

Re: Really?

[houghi]

People accuse others of rape when they become pregnant while it was not rape but sex without a condom. Stupid for BOTH, but not rape.
Somebody claims to have fallen asleep when they tatood her face because her dad was angry with the tattoo.
People claim somebody else ate the cookies while their face is full of crumbles (ok, not so much people, but toddlers)

We see cars accelerate a lot of the times and as far as I know it always is an automatic (hence it happens more in the US). So it does not seem unreasonable.

Few years ago a girl I knew n Germany was parking daddies mercedes and wanted to brake. Instead she pushed the accelorator a little bit. In panick she wanted to sslow down and that means push down you foot. You go faster? Push harder. As the foot was still on the accelerator, 10 cars were damaged.

In a car with shift, to go slower means to downshift as well at some point and that means moving your foot up and thus releasing the brake or in this case the accelerator.

To go from what you do as instinct to realize what is going on and correct that will take 2-3 seconds? And that is not even from the moment you start accelerating. So I can imagine this going like:
Press the wrong pedal. -Oh, I need to slow don-
Press the wrong pedal harder -I seriously need to slow dow--SPLAT--
If there is enough space it would go on with
Wait, I am not slowing down --notices mistake--
OK, Now I have the correct pedal, SIGH. Luckily nobody will EVER know about this.
So what I can see here is a normal automated reaction of a human that had the wrong result.

Re:Really?

[AcquaCow]

Not sure how their DBW system is setup, but even in my old 2004 Subaru, the DBW system uses two sensors and duplicate wiring.

There are two potentiometers in the throttle pedal and in the throttle body. A change in pedal voltage, reflects as a change in throttle body stepper motor voltage = open throttle body (to some percentage equal to the pedal movement).

If the two potentiometers are off by more than 0.05v, then the ECU shuts off the input from the pedal and the car goes into limp-mode where it won't open the throttle plate more than a few deg and won't let the car accelerate any faster than idle.

Re: Really?

[Immerman]

Truckers are (generally) some the the safest drivers on the road. Probably something to do with tests for commercial licenses being more rigorous, and their livelihood depending on not being found at fault for an accident.

Re:Really?

[kilfarsnar]

There's nothing preventing "45 year old wives" from knowing something about the machines they operate.

Really? Have you met many 45 year old wives?

Re:Really?

[internerdj]

Further, a manufacturer who may be at fault will always publicly interpret the user input sensor as user input.

Re:Really?

[Immerman]

Or might very well use the exact same wording. "On autopilot" = "on its own/without my orders", the fact that the car does or does not also have a feature called "autopilot" could be completely beside the point.

Re:Really?

[Immerman]

They do if they (think) they're pressing the brake and still not stopping. The automatic response is to push harder until you slow down at the desired rate. Making the connection that the "brake pedal" is actually causing acceleration and switching to the proper pedal is a higher-level brain function that happens much more slowly.

Re: Really?

[thegarbz]

If you're building a safety critical system, having two sensors won't do you much good. Even if they disagree, you have no way to know which is wrong, only that some fault exists. The best you can do in that situation is try to fail to safe

Almost. In the design of a safety critical system a 1oo2 option has a lower failure rate than a 2oo3 option if the safe state is known, and if the accidental triggering of the safety function isn't unsafe. Unfortunately for a car the safe state is not always known. Take a car in an intersection with an impending sideways collision, the safe state is not to assume acceleration is zero when the input becomes indeterminate.

Classic control theory for safe control is to take the media value (not average) of 3 sensors, and trigger an alarm if the sensors deviate by a certain percentage. That way during failure you're guaranteed to always have an error less than half of the deviation alarm (assuming the system is maintained and repaired when required).

Re: Really?

[thegarbz]

The worst case for the former is that your car doesn't accelerate. The worst case for the latter is that your car slams on the brakes. In neither case would you be at fault

No one cares who's at fault. Especially not the person in the morgue. A car accelerator problem has no safe state. The control action to avoid an incident changes depending on the situation. Something as simple as crossing an intersection will change the safe action from slamming on the break to slamming on the accelerator half way through.

This system can only be rendered safe with a mid out of 3 control scheme, the cost of which is trivial to implement.

Re: Really?

[thegarbz]

Mechanical linkages can be grabbed and wiggled to be evaluated. Nothing will tell you if an electrical component is about to fail.

You got that backwards. Mechanical linkages can not be monitored online and only checked in specific maintenance periods. Electronic components can monitored and diagnosed online line, compared with in logic, and be error corrected in ways that during component failure the worst the user experiences is a little alert that their car needs to be looked at.

It's all about the design you apply.

Re:Really?

[ArsenneLupin]

Your argument is valid but "his 45 year old wife behind the wheel" gets priority .

Well that still doesn't rule out the "or indeed anything else like a dropped handbag" theory :-)

Re: Really?

[Type44Q]

I bet you're a hoot at parties.

Re: Really?

[Rei]

How often does the average person "grab and wiggle" the mechanical linkages on their accelerator and brakes and evaluate whether they're likely to fail soon? A drive-by-wire system going into failsafe will let you know and will, as the name says, "fail safe" (the lower of two acceleration values or the higher of two braking values). A broken or stuck mechanical linkage will just suddenly and instantaneously be broken or stuck, and not inherently in a failsafe position. Stuck accelerators are a real and terrifying situation that does sometimes occur.

Re:Really?

[Opportunist]

Then I consider myself lucky that I don't have to drive in the US. Though it does explain why blunders like this can happen.

Re: Really?

[0100010001010011]

My 1998 VW TDI had a 2 sensor setup for the accelerator. If I remember correctly they were inversely linear resistors. 0% pedal was 0.5V on one and 4.5V on the other, 100% throttle was 4.5V/0.5V (respectively).

If they disagreed it did nothing. Yes, it left your car dead by the side of the road but that is the proper failure case for an automobile.

Re:Really?

[bluefoxlucid]

The likelihood of a rheostat failing in that way is incredibly low. The sensor in the accelerator is of such a design that a fault as such would require action (it wouldn't just *happen* because you shook it a little or whatever), and would consistently occur in the same manner.

Accelerator pedals in drive-by-wire systems are universally wire-wound potentiometers. They work by the physical movement of a connection arm along a linear or circular path, thus reducing the length of electrical travel across a resistive medium. Getting to 100% means traveling a full distance. If it were physically broken in such a way that that could just happen, the pedal wouldn't work at all; if it had a short spot on it, that would trigger on 100% of uses, and only at a certain position.

We're talking about something that literally has never failed in this way in billions of cars worldwide. They're built in a way which doesn't allow that, in the same way they're built in a way such that they don't apply torque and kick back. It's the same object as the volume knob on a guitar amp or old stereo.

Re:Really?

[bluefoxlucid]

Making the connection that the "brake pedal" is actually causing acceleration and switching to the proper pedal is a higher-level brain function that happens much more slowly.

It happens before my car starts moving. If I hear the engine start to growl or feel any form of torque upon depressing a pedal, I know what just happened. The accelerator is also a much lower-force pedal than the brake. There is so much instant feedback telling me what I just did I'd have to have part of my brain physically removed to not know wtf just happened. I'm pretty sure my autonomous nervous system reacts before my prefrontal cortex processes the information, too.

Re: Really?

[drinkypoo]

How often does the average person "grab and wiggle" the mechanical linkages on their accelerator and brakes and evaluate whether they're likely to fail soon?

A person with no mechanical sympathy is fucked. For the rest of us, you can feel parts failing in most cases. You might not be able to tell what is failing, but you can tell that something is getting sloppy, and that you need to give some attention to the vehicle. Sometimes I can pin the problem down just to a specific bushing or ball joint, but that's hard on modern vehicles with multilink front and rear.

A broken or stuck mechanical linkage will just suddenly and instantaneously be broken or stuck,

This almost never happens.

and not inherently in a failsafe position.

This depends on your design. If your throttle cable snaps, and the return spring is on the throttle body, then that's inherently failsafe. Suspension can even be designed this way; the suspension on my 1997 Audi A8 is pretty sketchy, if pretty much any link fails you're screwed. But the suspension on my 1982 MBZ 300SD is actually quite good. If any one link in the front suspension fails, things are supposed to fall in such a way that the car doesn't fall apart. Obviously, if you lose the link between the pitman arm and the steering rack, you're going to lose steering, but so what? There's a similar point of failure in every steering system.

So actually, you are spectacularly wrong about the first point since mechanical linkages almost always give warning before failure, and you are at least partly wrong about the second.

Stuck accelerators are a real and terrifying situation that does sometimes occur.

Yes, I've had it happen to me when a return spring broke on a 1971 Dodge Sportsman. It was the old, bad, sad design, with a big porch door spring next to a carburetor. That doesn't invalidate the concept of physical linkage. It invalidates the concept of poorly designed throttles.

Re:Really?

[guruevi]

You also have to look at HOW it got to 100%. If a sensor fails, the value will jump from let's say 10% (to park) to 100%. The log would show something like this: 9-12-10-8-9-12-100. If a user presses the pedal it goes something like this: 9-12-10-8-15-40-55-72-91-100 (given you measure something like this in human muscle reaction range which you probably do because otherwise the car would respond sluggish)

Re: Really?

[beelsebob]

But they also weigh something. Making triple redundant systems for every safety critical feature of the car adds a lot of weight through death by a thousand cuts. Weight makes a worse car.

Re:Really?

[beelsebob]

Yes - Toyota went for years claiming "oh, our cars are mostly driven by 60 year old women who don't know what they're doing" before an example became apparent where the police could prove unequivocally that the driver was doing everything they could to stop the car, while it was convinced it wanted to go as fast as it possibly could.

Re:Really?

[bluefoxlucid]

Brakes are generally heavy and require much more force than accelerators. Clutches require much more travel distance. If you did, somehow, miss the brake and hit the accelerator, you'd probably hit the accelerator hard enough to floor it--or just enough to nudge the brake gently to engaging.

Re:Really?

[bigpat]

Technology is about to be mandated that will stop incidents like this; mandatory Automated Emergency Braking is coming in what, 2020? 2025? Real soon now, since all the major manufacturers and suppliers have an AEB solution available. And then "wrong pedal" collisions will be a thing of the past. The car simply won't let you do that.

It's too bad enough of us couldn't be responsible for humans to be able to keep the responsibility of driving, but more and more of it is being taken away.

That is the best fail safe. I don't see any *good* reason a car should allow a driver to hit another object.

Re: Really?

[goose-incarnated]

Mechanical linkages can be grabbed and wiggled to be evaluated. Nothing will tell you if an electrical component is about to fail.

You got that backwards.

Nope; mechanical linkages give plenty of warning before breaking. Cables stretch, pedals have a different "feel", etc. For brakes, I've never heard of the mechanical linkage (a stout piston connected to the pedal) failing, but as it's a piece of metal, it's not going to suddenly snap. Metal doesn't do that.

Re:Really?

[jimbob6]

Hey some one got it right.
The log showing 100% acceleration is in fact evidence that there was some kind of mechanical fault. Not evidence of user error.
I don't know many users that press the pedal all the way to the floor even when there not running into a brick wall.
The inertia alone would cause the passenger to slide away from the pedal.

Re:Really?

[nrjyzerbuny]

If I hear the engine start to growl or feel any form of torque upon depressing a pedal, I know what just happened. The accelerator is also a much lower-force pedal than the brake. There is so much instant feedback telling me what I just did I'd have to have part of my brain physically removed to not know wtf just happened. I'm pretty sure my autonomous nervous system reacts before my prefrontal cortex processes the information, too.

But it's a Tesla, so there won't be an engine growl, and there may not be any feeling of spooling up torque. Electric motors are quiet and go from zero to Holy Shit in no time. It's entirely possible the driver began depressing the 'brake' (actually accelerator), had time to notice not slowing down, and in a panic pressed harder and put it to the floor.

I can't assert that's what happened of course, but it happens /all the time/ in cars that make more noise when accelerating, don't accelerate as fast, and have less power. There was a hole in the front of my gym from someone doing it this last winter, and it was the second time that person had done it!

Re:Really?

[Jeremi]

There is so much instant feedback telling me what I just did I'd have to have part of my brain physically removed to not know wtf just happened. I'm pretty sure my autonomous nervous system reacts before my prefrontal cortex processes the information, too.

It sounds like you are speculating about how you imagine it would be, rather than speaking from experience. The problem with that is that you're "primed" -- you've already been thinking about the issue while reading this discussion, so in your imagined scenario you already know what the problem is and how to fix it, and don't have to spend any (imagined) time figuring it out while under stressful circumstances.

I have had this problem happen to me, so I can tell you how it actually was: I had borrowed a friend's automatic-shift pickup truck to move some large items, and after putting the items into storage I got back into the truck to drive it back to him.

The truck's pedals were smaller and closer together than on the car I typically drove at the time, and so when I put the truck into reverse and then pressed down on the brake (to moderate the speed at which the car backed out of the parking space), the corner of my shoe caught on the edge of the accelerator pedal as well. This caused the engine to race and the car to suddenly accelerate backwards (since the brake pedal's travel was larger than the accelerator's), and my automatic (muscle-memory) reaction to correct this was to press down on the brake harder, which only caused the car to move faster.

From memory, it took me about half a second to figure out what might be actually going on and lift and then re-center my foot on the brake to stop the truck. That was just barely enough time to avoid slamming into the brick wall on the other side of the (rather small) parking lot. That was when I was younger and at a moment when I wasn't fatigued or distracted -- I'm not sure that most people (or even myself, anymore) would necessarily be able to diagnose and fix the problem that quickly, if they had never experienced it before and weren't expecting it.

There is so much instant feedback telling me what I just did I'd have to have part of my brain physically removed to not know wtf just happened.

"Having part of one's brain removed" is a pretty good description of what happens when a person panics -- the intellectual part of their brain shuts down and the low-level flight-or-fight subsystem takes over. Handy for escaping from lions, perhaps, but our reptilian brains aren't very good at diagnosing automotive control problems. :(

Re:Really?

[operagost]

What are you talking about? Most, if not all, states in the USA have a points system. If you are involved in repeated incidents (moving violations, accidents), you'll have your license suspended and sometimes be required to take a course to get it back.

Re:Really?

[ripvlan]

yes and - where was the Collision Avoidance system in all of this? I understand that maybe it can't see the back of a flatbed truck... but a Wall?

But I suggest that these analyses receive independent review. We are all willing to take the Tesla word for it as final.

Re:Really?

[beanpoppa]

I did have an issue with an Audi several years ago. My cruise control would periodically turn itself on. They checked the computer, and it was reading occasional faults in the cruise control switch, so they replaced it. The problem continued. Sometimes over bumps, but also sometimes when I would turn on my turn signal. The turn signal and cruise control switches were both stalks on the left side of the column. The dealer read the codes again, and again said the cruise control switch was reporting faults and wanted to change it again. When I protested, and pointed out that it would turn on when I turned on my turn signal, they assumed that I was mistaking the two stalks. (to which point I asked them to investigate why my turn signal was turning on when I activated my cruise control...) It was only after the service manager finally driving it around and being able to duplicate the issue that they believed me. They ultimately replaced the control module in the steering column, which contained both switches, which solved the problem. Not saying that the brakes and gas pedal go into the same control module, but I can certainly see a case where the control module has a fault which is reading the wrong input.

Re: Really?

[budgenator]

Yes and they also give you tactile feedback, so you can learn what feels right and what feels wrong as you operate them.

Re:Really?

[PinkyGigglebrain]

I've got two questions about the log:

how many times a second does the accelerator sensor get sampled?

Does the log show anything between 6% and 100%?

If the sensor went from 6% to 100% in 100th of a second with no intermediate values being recorded I would have to wonder if the sensor failed.

Re: Really?

[Rei]

A person with no mechanical sympathy is fucked.

Then the vast majority of the population is fucked. Great plan.

This almost never happens.

A mechanical linkage is either attached or not attached.

That depends on your design

Yes, depending on the failure scenario, it might or might not get stuck in an on or off position. With a single-point failure in a dual-sensor electronic linkage, it's always failsafe (and single-point failures always detected since the reported values disagrees or is absent). Only simultaneous failures can lead to a non-failsafe situation. And even in that case, that only brings you to a "maybe" situation, depending on the scenario. For example, simultanous failures may simultaneously fail to 0; they may simultaneously fail to no signal; etc. In those cases, you still get failsafe.

Re: Really?

[stealth_finger]

- (automatic): The A320 has independent Angle Of Attack probes, but if you climb through heavy icing, it turns out two can freeze up at the same time. The system sees a confirmed stall (even though it should be impossible given the current speed, attitude and inertial measurements) because two systems agreeing can't ever be wrong, and it pushes the nose down. Even if both pilots pull back on the stick, the nose continues to drop. We now have an official procedure to shut off two of the Air Data Reference units, leaving only one remaining. That's the only way of shaking the absolute confidence of the system in two identical but false signals.

I'm not sure that's accurate. It's slightly different situation but that air france 447 was an airbus, and its sensors froze causing the pilots to think it was stalling so one of the pilots for some reason decided to pull up and it stayed pulling up until it hit the deck. Maybe your thing was implemented after that though?

Re:Really?

[Anonymous Brave Guy]

The control systems for a lot of modern cars are far more interdependent than they really ought to be, despite obvious (to someone with the right background) safety and reliability implications. What you wrote there is plausible, but I certainly wouldn't just take a report from the manufacturer at face value in this sort of case. Auto manufacturers don't exactly have a reputation for being honest about these things, particularly when business-destroying liabilities or very expensive safety recalls are on the cards.

Re: Really?

[Anonymous Brave Guy]

Majority voting with three systems has turned out to be dangerous on quite a few occasions.

Any fault in these kinds of systems can be highly dangerous and should be grounds for emergency procedures and taking the equipment out of service as soon as possible. That's as true if you have majority-of-three as it is with just two readings. Systems should still be designed to fail as safely as possible under those conditions, which might still mean assuming the single reading to be the correct one if that would be the safest response under the circumstances.

The problem with both examples you gave is that the emergency procedures did not err on the side of caution and solely trusted the majority equipment. Even if you've got a heavily automated system like a modern Airbus, it is clearly a concern if both pilots can indicate that they want the plane to do something and yet the control systems decide to do something else anyway.

Re:Really?

[Cederic]

how many women do you know who could actually change a tire?

I think 2/3 to 3/4 of the women I know could actually change a tyre. Maybe 1/20 would - the rest would ring the AA or the RAC.

Pretty comparable to the men I know, on the whole.

Don't forget that a lot of people lack the physical strength to loosen a wheel nut that's been overtightened by a zealous garage worker intent that he wont be responsible for your wheel accidentally falling off.

Re: Really?

[stealth_finger]

Would you really wanna be in the middle lane of a 5-lane, 65 mph, highway and have your car 'controlled brake' to a stop? What about when your pulling out of a parking lot, turning left across a busy 2 lane road? Crossing railroad tracks? On an icy road? There are quite a few situations where I wouldn't want my car to begin a controlled braking action. I would tend to agree that there is no 'perfect' failsafe action that does not take into consideration the context of the situation.

Would you want it to apply full acceleration instead? It seems to me in all those scenarios it would be better for the car to do nothing with the throttle and let it coast to a stop.

Re:Really?

[epyT-R]

More like "too bad we let our feels get in the way rather than let darwin weed out the incompetent." Preferably, this would happen long before driving age.

Re:Really?

[ilsaloving]

...or you could live in the real world and acknowledge that people don't buy cars to learn about them but to drive them.

Then they shouldn't be buying the damn car.

Too many people seem to think that driving is a right. It's not. It is a privilege that can and should be taken away if you demonstrate that you cannot be bothered to take due care while operating a a large and complex piece of machinery.

Re:Really?

[stealth_finger]

Most of all I don't use more words than absolutely, positively and ultimately required to express the message that I wish to convey.

So you are right. The word I should have used is "demand".

Good luck with that one.

Re:Really?

[stealth_finger]

Most of all I don't use more words than absolutely, positively and ultimately required to express the message that I wish to convey.

So you are right. The word I should have used is "demand".

Good luck with that one.

Also I would put it to you that positively and ultimately are unneeded to express the message you are conveying. I would assume sarcasm but you seem to be really quite serious, so not so sure. Are you sure, absolutely sure that you couldn't have expressed that message any shorter?

Re:Really?

[Solandri]

We went through this in the 2000s with Toyotas and in the 1980s with Audis. I hear it happened earlier with Subarus (i was too young then). They lawyers call it unintended acceleration, to try to make it sound like the driver isn't to blame.

The auto manufacturers did a ton of research on it, and concluded it was pedal misapplication - people pressing on the accelerator when they thought they were on the brake. It's an innocent mistake (I've done it myself a couple times), but some drivers think they are perfect and refuse to admit they made a mistake, and hire lawyers to sue the car manufacturers. You have these people to thank for the gearshift interlock which prevents you from shifting out of Park unless you first press down the brake pedal. After Audi installed that, all the mysterious cases of "unintended acceleration" disappeared, and all the other car manufacturers added the same "feature" to protect themselves from stupid lawsuits.

In these unintended acceleration stories, the driver always says they were pressing as hard as they could on the brakes. That's because if they weren't doing that, people would blame them for not doing everything they could to stop the car. Here's the thing though. Unless your brakes have failed, your accelerator can never overpower your brakes. On a ICE vehicle, the brakes are simply more powerful - they can generate a bit over 1 g of stopping deceleration, while the engine can only produce about a half g of acceleration (0-60 mph in 5.5 seconds). Electric motors could in theory overpower the brakes, but only the Tesla S in ludicrous mode can exceed 1 g of acceleration (1 g is 0-60 mph in about 2.7 seconds). On hybrid/electric vehicles, there's usually a cutoff which prevents power from going to the motors if the brake is depressed.

So even if the accelerator sensor (which due to the serious nature of 100% uncommanded acceleration doubtless was designed to fail low and tell the car there's 0% depression if something goes wrong) gets stuck at 100%, you can overpower it by stomping on the brakes. And cars have 3 redundant and independent braking systems - hydraulic and mechanical linkages to the foot pedal, and an independent parking brake. So for these stories of unintended acceleration to be true, the sensor would have to get stuck at 100%, and all 3 independent braking systems (5 for hybrid/electrics - they add the aforementioned cutoff and regenerative braking) would have to fail simultaneously during the event, then all 4 (6) systems would have to mysteriously begin working properly again during later testing.

While the odds of this happening aren't zero, they are exceedingly small. It is much much much more likely the guy stomped on the accelerator when he meant to hit the brake.

Re: Really?

[Khyber]

"but as it's a piece of metal, it's not going to suddenly snap. Metal doesn't do that."

That depends ENTIRELY on the manufacture and grain direction of the metal crystalline structure.

I've snapped plenty of leather punch-awls with simple pressure.

Re:Really?

[bluefoxlucid]

and there may not be any feeling of spooling up torque

This is the feeling of the car--or, more precisely, the seat your ass and back are in (or suddenly applying force into)--moving and applying force against you. If there is no feeling of torque kicking in on the wheels, there is no acceleration.

The pedal is still softer. Accelerators can generally go down with 1g-5g; and modern one-pedal driving cars apply resistive braking when you let off the pedal, so braking would kick in immediately as the driver moved to the brake in response to touching a really soft pedal and getting kicked into her seat by insane amounts of instant torque.

There was a hole in the front of my gym from someone doing it this last winter, and it was the second time that person had done it!

Look, I don't sit around when something happens and go, "Huh? What's going on? Wait, I know this feeling. Why is this happen now? This shouldn't be happening now! OH SHIT!" If I did, I'd have third-degree burns from repeatedly grabbing the handles of smoking-hot cast iron frying pans (I don't learn very well from negative consequences; although psychologists claim positive punishment doesn't extinguish behavior, and so the avoidance of burns by immediately withdrawing my hand when I touch a 700-degree iron handle is an effective a way of removing the negative consequence while also frequently allowing me to grab an only-warm pan, so maybe I'm just optimized to not waste time fetching a pot holder until I discover a hot handle).

I assume other people reflexively respond to familiar and unwanted stimulus by taking mitigating actions they've learned through thousands of repetitions. You've hit the accelerator hundreds of thousands of times, and encountered the perception of movement; you've hit the brake thousands of times; if you hit the accelerator, you should recognize that, and know how to undo it.

To be fair, I did fail to handle a new situation recently. A cat ran across a highway in front of me. Going 50mph and not having registered that there were no cars behind me, I used the normal strategy of partial braking while checking my mirrors; I did not stop fast enough. Maximum braking force would have inflicted minor injury on me and my passenger (my car can stop with ~10-12 linear Gs; it's like hitting a tree), but nothing serious. It took me between 2 and 3 real-time seconds to assess the situation (the subjective memory was *considerably* distorted and projected ~10-15 seconds; this required correction), meaning I was physically and mentally capable of stopping in around 1/3 that distance *if* my brain would have selected for maximum braking force with the same reaction speed.

I've replayed that event several hundred times since, and removed all the psychiatric trauma; the new responses are only weakly embedded, so good luck if your kid walks out from behind a parked van 4 feet in front of me in a 30mph speed zone. My normal manner of driving involves scanning the area ahead, including sidewalks, to account for pedestrians and other independent behaviors; the cat exercised an unpredicted behavior before I recognized there was a cat, so that situation should not frequently translate to unleashed humans. I think.

None of this would matter if my brain couldn't select from a set of known behaviors based on a set of recognized conditions *before* informing me wtf is going on and why I'm doing something. Without that, I'd just crash into things all the time. I assumed everyone else had similar facilities, since it's a core part of human neurology.

Re:Really?

[Khyber]

If a dropped hangbag pressed the pedal, it would've had to be filled with a few very heavy objects - like bricks - to make it move enough to register 100% throttle.

Pretty much every pedal I've had my foot upon needs several pounds of pressure to get it to maximum depression.

Re:Really?

[Khyber]

No luck needed. You don't learn this vehicle, you don't get to purchase or drive this vehicle. Make it law. Suddenly you'll see our overall population intelligence go up, or you watch as the idiots in this country realize what hole they dug themselves into with their own stupidity and ignorance, while the rest of us drive on by.

Re:Really?

[Khyber]

I only know three or four women 45+ years old that can do any real mechanical work.

That changes when you get into the 70+ year old range - they're a bit better because they came from a generation where you fixed things and did things yourself.

I know maybe two younger girls (21+) that work on cars - HOWEVER they're only into the mechanical stuff, nothing electrical excepting the alternator and wiring harness interests them in that regard.

And I know WAY more girls than I do guys - the eternal gay curse.

Re:Really?

[Khyber]

" I don't see any *good* reason a car should allow a driver to hit another object."

Then you've probably never been shot at while driving your vehicle. Try living in Memphis, some time.

Re:Really?

[bluefoxlucid]

It sounds like you are speculating about how you imagine it would be, rather than speaking from experience

No, this is how I actually drive. I *often* have no idea what the hell's going on until after I've done something. I frequently grab hot objects without burning my hand because I react to handling a 700 degree piece of metal by removing my hand from it. On the highway, I usually *predict* what's happening around me; but, when surprised by something I hadn't predicted at all, I immediately respond--often wrongly, by suddenly braking, because braking is a known response; I've done lane tosses frequently enough (instead of braking, you accelerate into an opening to go around an obstacle, which is safer), but not with perfect reliability.

I have *also* slipped my foot off the brake and floored my accelerator while behind a parked car. I didn't hit the car. The first time I did this, I spent quite a while considering htf something like that happens.

This caused the engine to race and the car to suddenly accelerate backwards (since the brake pedal's travel was larger than the accelerator's), and my automatic (muscle-memory) reaction to correct this was to press down on the brake harder, which only caused the car to move faster.

See, when the acceleration kicks in, my brain goes, "Oh you must be on the accelerator!" The brainstem picks up on that as the midbrain feeds it sensory information (sound, visual cues, the feeling of my body as the car moves in an unexpected direction) faster than it gets to my prefrontal cortex. It then issues commands before I have a chance to assess and react to the situation. My body does the right thing.

"Having part of one's brain removed" is a pretty good description of what happens when a person panics -- the intellectual part of their brain shuts down and the low-level flight-or-fight subsystem takes over.

This has only happened to me during a psychotic episode. Particularly, it's only happened when I start hallucinating severely at night, alone, in the dark, due to nyctophobia triggered by being alone, in the dark, at night. My visual cortex projects an internal visualization of the current visual field and then alters it; the rest of my brain responds to the altered version, as it coincides with the original aside from a few projected specters. I'm capable of walking around anywhere blindfolded because of this.

In any case, my brain responds to stimulus with an appropriate reaction based on very blunt reasoning. My response-inhibition system requires the involvement of my prefrontal cortex; but "I want X and action A will get X" is what you'd get if you removed my entire frontal lobe, and it's enough for my brain to reason that the feeling of movement is incorrect and can be corrected by turning the steering wheel or pressing the brake or whatever is required.

Re:Really?

[benro03]

I doubt that it's just a sensor reading acceleration, especially modern autos use "fly by wire" accelerators and not physical linkages anymore.
If I were the obsessive compulsive engineer that Elon Musk is (and seems to hire engineers that way), I would not only have a sensor recording the acceleration I would have a pressure sensor on the pedal itself. That way, not only does it read the call suddenly accelerating to 50 mph, it records the 50+ pounds of pressure on the pedal.

Re: Really?

[mrthoughtful]

I'm a bit confused by tenses when it comes to fictional historic narrative set in an imaginary future.
Wouldn't it be better to say "The Raman's will have been having always done things in three's and look how that will have been turned out (though it could have been the fact that they will have been having used left-handed threads, for all we know)".

Re: Really?

[kcwebmonkey]

Majority voting with three systems has turned out to be dangerous .

Well obviously 3 parallel systems won't work. We need to use a tried and true system... like superdelegates!

Re:Really?

[CanadianMacFan]

I know one who is one of the best software developers that I know and having working for a couple of decades I've come across many in my city.

Re:Really?

[rahvin112]

Color me a skeptic as well, love Tesla would love to own one but Toyota made the same fucking claims about their unintended acceleration that sent people to jail.

Re:Really?

[CrashNBrn]

So people that are "45" didn't grow up with the Internet? Fuck you. Clueless idiot. Learn history.

Re:Really?

[stealth_finger]

So what is it you propose? A one time check to make sure people can operate a vehicle safely? A periodical such? That one has intimate knowledge of internal combustion engines? Be able to strip the car to it's component parts and put it back together? Something else? Personally I'd say you'd have to renew your licence every ten years as a good compromise. Where do you come down?

Re:Really?

[F.Ultra]

Such accelerations is not unheard of yes but have there even been a single case where it has been proven to be the fault of the car? And do not 99% of these cases always comes from the US? A country with a judicial system which encourages fraud like this.

Re: Really?

[AaronW]

Out of all the cars I've driven I've had two that have had stuck accelerators. Both had mechanical linkage. Give me drive by wire any day.

Re:Really?

[l0n3s0m3phr34k]

Speak for yourself, my driving style involves smashing the accelerator into the floor at all times, ESPECIALLY in reverse.

Re: Really?

[AaronW]

I had a car where it stuck (1991 Ford Probe) and no, you could not feel it through the pedal before or even when it happened.

Re:Really?

[stooo]

>> but if the logs show 100% acceleration, that just reflects the sensor value. Not that the user - or indeed anything else like a dropped handbag - actually pressed the pedal that far.

Acceleration sensors have redundancy with consistency checks
If a sensor fails, it is logged. And this applies to every modern car.

Re:Really?

[stooo]

45 year old Wives, on the other side, usually don't have consistency checks.

Re:Really?

[AaronW]

In the case of Toyota while many cases could probably be attributed to a problem between the steering wheel and the seat, it was also found that the ECM code was horribly written spaghetti code that had numerous flaws in it that didn't follow most of the guidelines for automotive design.

In this case since they had only had the car for 5 days I say it was a problem between the steering wheel and seat.

Re: Really?

[Nahor]

If two out of three can be wrong, why can't two out of two be wrong? If you have a fail-safe, you should always do it if even one sensor is off, regardless of the majority

But if you don't have a fail-safe, or if the fail-safe is too costly, or if the sensors are not that reliable, with 3 sensors you have more chances to make the right decision by following the two sensors that agree. With with only two sensors, you'll have to follow one of them, i.e. you have a 50-50 chance of making the wrong choice.

Re: Really?

[Anonymous Brave Guy]

That's not a safe assumption at all. Killing the driver's intended acceleration while traversing an intersection or merging with a high speed road would be extremely dangerous, for example.

This is a serious problem when designing safety systems related to cars: in some cases, there is no universally correct safe mode you can fail to if a fault is detected.

Re: Really?

[F.Ultra]

If so then surely the logs would contain rows like "enable autopilot." and "autopilot disengaged due to strange error". Also it's not like the car is totalled into a 2x2m cube so testing can be done on it.

Re: Really?

[stooo]

Today, a lot of them go to magnetic sensing (hall or GMR). That eliminates faulty pots.

Re:Really?

[soapdude]

Not the majority of them at least. A very small minority of people knew anything about the internet pre-1994, when 45-year-olds would have been 23. Therefore the overwhelming majority of them would have learned something about it after the age of 23, which hardly means "grew up with".

Re:Really?

[Anonymous Brave Guy]

Proven in court and caused specifically by the electronic control systems? Not that I'm aware of, though there have certainly been sudden acceleration incidents that weren't driver error and related attempts to cover them up, and there have also been cases demonstrating an ability to gain remote control over electronic control systems affecting vehicle movement.

However, as I noted before, there are also plausible explanations involving driver error in a lot of these cases.

Re:Really?

[budgenator]

I work with several 45ish year old women and operating machines at their optimum potential just isn't one of their emotional values. They tend to be more people orientated than technically orientated, hell if something doesn't break for me to fix, they'll just call me to come up and say hi every once in a while. That doesn't make them either better or lesser than anyone else, their strengths are a valuable addition to our team.

Re: Really?

[thegarbz]

Cables stretch, pedals have a different "feel", etc.

Ahh yes, the old feeling that something isn't quite right somewhere. I'm going to put that one down as the least reliable detection mechanism of them all.

For brakes, I've never heard

Stopped reading right there since I was in a car which had breaks fail for no reason attributable to wear while I was in it. The car was 9 months old.

Anyway all of this is beside the point. You said "nope" without refuting the core point. Electrical failures are measurable, detectable, and can be engineered around with ease. Mechanical linkages and mechanical systems while they are capable of being made redundant become insanely expensive whereas electrical redundancy and diagnostics are a trivial component expense.

But hey if you think otherwise then Exida, TUV and the many hundreds of companies around the world which publish failure rate analysis from FMEDA studies as well as citing data used in service would love to hear how they are all getting so wrong. You'd have a nice career in the chemicals industry too, plant managers love being told that their old non-redundant mechanical equipment is actually perfectly safe and that they don't need to spend $trillions (across the industry) on upgrades.

it's a piece of metal, it's not going to suddenly snap. Metal doesn't do that.

And I see I'm not talking to a mechanical engineer. A fun fact is that a large portion of electrical failures are actually to do with the metal that carries the magic smoke.

Re:Really?

[PCM2]

Pretty much every pedal I've had my foot upon needs several pounds of pressure to get it to maximum depression.

Which I take to mean the point at which you decide to drive your car into a wall at top speed?

Re:Really?

[mink]

There were BBS systems and services like AOL before the internet that offered much of the same (at the time) service as what the internet provided starting in the 80's

Re: Really?

[thegarbz]

What you're describing is not a result of a a problem with a majority voting system but rather something known as a systematic failure. Majority voting system are designed to counteract random failures not systematic ones. The inherent assumption is that the 3 instruments are measuring the same thing at the same time with any differences between them being truly random with common failures either between all or none of the systems.

This is quite hard to design correctly if you have engineering constraints that prevent correct working, but none the less it's not a problem inherent in majority voting systems which are still the most reliable in terms of action and prevention of accidental action.

Re:Really?

[thegarbz]

It's too bad enough of us couldn't be responsible for humans

It's not about being responsible. Humans are fallible. They make mistakes unintended (kind of the definition of mistakes). There simply is no such thing as a perfect human.

Re:Really?

[HiThere]

There are reasons, but they are rare enough that there should be a requirement for a manual override control that automatically disengages after a minute, or possibly 30 seconds. And the control should need to be both depressed and twisted.

Re:Really?

[HiThere]

Sorry, but biological evolution is too slow to keep up with technological evolution. So that approach is guaranteed to fail.

Re:Really?

[bigpat]

" I don't see any *good* reason a car should allow a driver to hit another object."

Then you've probably never been shot at while driving your vehicle. Try living in Memphis, some time.

I am not saying people shouldn't be allowed to use their car as a deadly weapon, just that perhaps there should be a big red button to disengage safety controls that says "Allow me to hit stuff"

Re:Really?

[HiThere]

You were doing fine up until your last sentence. Then your ability to self-examine was called into question.

Re:Really?

[HiThere]

IIRC, in the case of Toyota, while the code may indeed have been poorly written, they decided the real problem had to do with the design of the floor mats. (I never followed up to find out exactly WHAT the design problem was.)

Re:Really?

[HiThere]

While your comment is certainly correct about interdependencies and reliability, in THIS case I tend to believe the logs.

OTOH, it would certainly be a good idea to have them audited by an independent investigator.

Re: Really?

[Ichijo]

I bet a self-driving car wouldn't hit a couch or a stopped Tesla.

Re: Really?

[goose-incarnated]

Cables stretch, pedals have a different "feel", etc.

Ahh yes, the old feeling that something isn't quite right somewhere. I'm going to put that one down as the least reliable detection mechanism of them all.

For brakes, I've never heard

Stopped reading right there since I was in a car which had breaks fail for no reason attributable to wear while I was in it. The car was 9 months old.

It's a pity you stopped reading; reading frequently cures ignorance. Those brakes that failed most certainly did not fail due to the piston snapping. Drive-by-wire systems for brakes still has the piston there, only now there are also electronics to fail as well.

We're talking about drive by wire systems for brakes (note carefully the spelling).

it's a piece of metal, it's not going to suddenly snap. Metal doesn't do that.

And I see I'm not talking to a mechanical engineer. A fun fact is that a large portion of electrical failures are actually to do with the metal that carries the magic smoke.

Metal doesn't suddenly snap. Metal's failure modes are, barring exceptions, gradual degradation. You would do well to look up 'malleability'.

Re:Really?

[mink]

Sometimes it's people trying to pull a scam though,

I remember that James Sikes guy who was bankrupt and owed an a$$-ton of money (around 700K ) to creditors and 19K behind on the loan on the Prius pulling something fishy.

I believe it failed and he had a history of insurance fraud, filing false police reports, theft and regular fraud that made his story rather suspect (also the story was unbelievable in almost every detail).

Re: Really?

[michelcolman]

No, that was a different situation. On Air France 447, the pitot tubes froze and resulted in incorrect speed indications, with stall warnings and overspeed warnings at the same time. Extremely confusing for the pilots. They ended up stalling the aircraft because they didn't know which information was accurate and which was not, even though the plane was actually perfectly flyable and the airspeed indications had returned to normal operation.

The problem I described is a different one, and is actually the opposite: here the plane is actually convinced it is stalling while it's perfectly obvious to the pilots that it is not. In this case, contrary to AF447, the pilots are right but the plane refuses to obey and pushes the nose down anyway. This has occurred a few times already but fortunately has not resulted in any crashes.

It just goes to show that it's often easy to blame the pilots ("the AF447 pilots should have trusted the stall warning and recovered by pitching down") while in reality the warnings can indeed be false and it's sometimes hard to tell what situation you are really in. Airbus used to claim that stall warnings could never be false, and the Alpha Floor protections would never be activated erroneously, until they were faced with the double AOA probe freezing incidents.

An AOA probe, by the way, is kind of like a weather vane but mounted on a horizontal axis to measure the angle between the airflow and the aircraft (angle of attack). A pilot tube is a horizontal tube with a hole in the front to measure dynamic pressure.

Re: Really?

[michelcolman]

With two sensors, you should definitely not follow one of them. You should use a contigency procedure that is safe in all circumstances. For example, with two altimeters, go with the lowest indication. With two gas pedal input values, go idle and turn the four blinkers on.

The problem with 3 sensors is that a discrepancy ought to be treated very carefully, taking the possibility into account that the odd one out might be the correct one, but in practice this is usually not the approach that is taken. And this can actually make a three sensor majority voting system more dangerous than a two sensor system that trusts neither and acts accordingly.

Re:Really?

[Anonymous Brave Guy]

I certainly hope they did. What I don't know is whether they would then be honest in reporting their findings. The auto industry has been caught blatantly, knowingly and repeatedly lying to everyone up to and including governments and regulatory bodies in the past, and with the huge fines or recall costs potentially at stake, they have every incentive to continue doing so until the rules are changed.

Re:Really?

[Anonymous Brave Guy]

Agreed on both points. It's entirely possible that the logs are correct, but I wouldn't trust that just on the auto company's say-so without independent verification.

Re:Really?

[dcw3]

That together with your user name make complete sense.

Re:Really?

[EndlessNameless]

I frequently grab hot objects without burning my hand because I react to handling a 700 degree piece of metal by removing my hand from it.

That's reflex, not conscious action. Most people do not have reflexes fast enough to avoid burns at that temperature; conscious movement takes significantly longer.

Unfortunately, there is no equivalent reflex for leg muscles while driving so your analogy is irrelevant.

See, when the acceleration kicks in, my brain goes, "Oh you must be on the accelerator!" The brainstem picks up on that as the midbrain feeds it sensory information

Going up to the midbrain is much slower than reflexes, which are aleady barely be adequate for the task.

Rolling into a parking space at 5-10 mph, you are moving about 10 ft/sec. Tactile reflexes are on the order of 150ms, so you've traveled at least 1.5 feet before you are physically capable of responding---and gained some speed as well.

On top of that, you still have to "decide" to switch pedals, move your foot to depress the brake, and slow to a halt.

Assuming your decision is essentially a reflex response (or a learned reflex) which does not require higher processing, you are still looking at 250ms before you could possibly start moving your foot. So we're looking at 2.5 feet of travel, plus whatever you might gain from acceleration.

If we generously assume you could reposition your foot from the gas to the brake in another 250ms, you've traveled at least 5 feet unexpectedly with the vehicle angling into a space. In most parking scenarios, that alone will be enough to cause a collision.

And that 5-feet of movement is an unrealistic best-case scenario---no need for conscious thought and practically superhuman movement from accelerator to brake. A real-world acceleration error would easily be 10-20 feet due to the delay for cognitive processing in someone who doesn't exhibit the reflex responses you claim to have. And again, this is not accounting for speed gained during that period of accidental acceleration.

Even under best-case assumptions, a collision is fairly likely. A vehicle would be virtually guaranteed a collision even under favorable real-world conditions.

Re:Really?

[dcw3]

Try to keep up. You're talking about after the fact. And even then people don't give a shit about learning how the fuck they should be driving. There is no legal requirement for them to know, or be tested upon their knowledge other than the silly exams that just about any preteen could pass. Do you think more than a small percentage of them have a fucking clue? Drivers licenses are handed out like candy in the US, and it shows by how poorly people drive....sitting in the fast lane, talking on their cell phones, putting on makeup, reading papers. Does any of that matter to them? Do a few points matter to them? Not until they're close to losing their license.

Re:Really?

[EndlessNameless]

At the very least I expect someone to be able to operate a machine so that it is of no harm to me.

Even experts make mistakes.

As I suspect the 40-year-old female driver already knows how to operate standard automobiles, the accidental acceleration was a mistake rather than negligence.

the person has no right to operate the machine

There is a huge population of people who are competent in the operation of a machine yet are still mentally and physically capable of erring.

I've never accelerated when I meant to brake, but I have activated the turn signal instead of the cruise control on several occasions.

Re:Really?

[dcw3]

Ah, then I consider you lucky as well. The only place I've driven extensively, where people had a clue, was Germany(six years back in the 80-90s). Yes, they drive fast, but they always pass on the left, always use their turn signals, and it's extremely rare that someone does something unexpected, though it got a bit sketchy when the East Germans started showing up. I remember being told that drivers training cost the equivalent of about $1500 back in the '80s.

Re:Really?

[dcw3]

Rolling into a parking space at 5-10 mph, you are moving about 10 ft/sec.

5-10mph? Maybe straight in, and slamming on the brakes, but parallel?..not a fucking chance. 2 or maybe 3mph if you're in a hurry.

Re:Really?

[EndlessNameless]

but if the logs show 100% acceleration, that just reflects the sensor value. Not that the user - or indeed anything else like a dropped handbag - actually pressed the pedal that far.

There are actually two data points worth considering. First, the depression of the gas pedal to 100% was either a real event or a sensor failure---autopilot does not register pedal activation. But also worth consideration is the absence of brake activation.

When parking, normally the brake will be depressed as the car rolls into the space. Even if the car accelerated without any input, you would expect panic braking in response. So even if the gas pedal erroneously registered 100% depression, you should see gas and brake depression simultaneously.

A lack of simultaneous gas and brake activation would point toward a user error followed by a corrective response that was, unfortunately, too late to avoid a collision.

Re:Really?

[Opportunist]

It's similar in many countries in central Europe. Germany, Denmark, Austria, Switzerland... Getting a driver's licence costs a small fortune (200-3000 bucks, MUCH more if you plan to get a license for trucks, and for really large ones you first need to drive smaller ones for a few years before you're even allowed to get a license), it's VERY easy to lose it (for a month at first, repeat offenses usually mean longer to forever), insurances are very happy to refuse payment if you cause an accident by stupidity (or rather, they pay to person you hit, then sue you for all you have and then some, and usually courts are on their side), and the laws governing road behavior have their own law book (size of a phone book). And you are expected to know and heed them ALL.

And lo and behold, people do it. Darwin's law at work, those who can't won't have their license for long. What you see on the roads is what's left after idiots, drunks and cholerics have been weeded out.

Re:Really?

[Khyber]

Prove you're capable of doing the most basic maintenance upon, and understand the basic operations of, the vehicle you're driving.

No need to tear down an engine, but you'd better at least damn-well know the ins and outs of the specific vehicle you wish to purchase.

I've seen people roll off the lot without ever figuring out how to operate basic shit like their headlights or turn signals.

Re: Really?

[David_Hart]

Would you really wanna be in the middle lane of a 5-lane, 65 mph, highway and have your car 'controlled brake' to a stop?
What about when your pulling out of a parking lot, turning left across a busy 2 lane road?
Crossing railroad tracks?
On an icy road?

There are quite a few situations where I wouldn't want my car to begin a controlled braking action.

I would tend to agree that there is no 'perfect' failsafe action that does not take into consideration the context of the situation.

Would you want it to apply full acceleration instead? It seems to me in all those scenarios it would be better for the car to do nothing with the throttle and let it coast to a stop.

Anyone in the Northeast can tell you that any type of breaking on black ice is BAD. The interesting part is that if you coast and slightly steer the car you can maneuver on black ice and slowly reduce your speed, but as soon as you start breaking or steer too hard you lose all control. I'm also thinking that black ice is going to be one of the more difficult situations for self-driving vehicles to detect.

If you are going through mud, for whatever reason, you do not want the car to auto-break. If it does, you'll likely end up stuck. Just ask the off-road drivers. While this isn't an immediate dangerous situation, you could end up stuck on a back road in the middle of nowhere because the automatic system decided that it needed to apply the brakes. Without that feature, you might have been able to plow through.

So, yes, there are conditions where you wouldn't want the automatic system to do anything. The hard part is being able to identify these situations and then selecting the right response. Personally, I'm willing let others debug the system for me... (grin)

Re: Really?

[thegarbz]

Then you'd do well to go back to the original problem which is that malleability is not detectable in a mechanical system, only by inspection.

So we're back to square one: An electrical system which can be setup to be reliable, redundant, self-diagnosing, and with predictable and controllable failure modes, vs .... a mechanical wire which needs periodic and highly unscientific inspection.

And by that I mean there's a lot of room for judgement in the inspection of mechanical failure modes, not the kind of system that you want to entrust to an industry mostly consisting of people who didn't finish school.

So to summarise around the pedantry: You want redundant electrical fly-by-wire systems in a car if you're going for reliability and safety. Hell you want that in most of industry too.

Re: Really?

[goose-incarnated]

Then you'd do well to go back to the original problem which is that malleability is not detectable in a mechanical system, only by inspection.

Nope, wrong again - malleability affects the dimensions of the part. Dimensions are something that can be felt just through usage, or heard due to varying clearances between parts. The part in question, which you originally chose to question, was the piston used to provide pressure to the hydraulic system in brakes. Should this part fail it degrades gradually.

So we're back to square one: An electrical system which can be setup to be reliable, redundant, self-diagnosing, and with predictable and controllable failure modes, vs .... a mechanical wire which needs periodic and highly unscientific inspection.

I did not dispute that, I disputed brakes, which you then took issue with. As far as I am aware, even the Teslas still have hydraulic brakes, in which case the non-fly-by-wire system is superior to a fly-by-wire system. And as for your "An electrical system which can be setup to be reliable, redundant, self-diagnosing, and with predictable and controllable failure modes", just because it *can* doesn't mean that it will. Mechanical systems can also be set up to be reliable, self-redundant, self-diagnosing with predictable and controllable failure modes.

In fact, they usually are. The hydraulic master cylinder is designed with two independent chambers that each serve two wheels that are diagonally opposite each other, so if a leak springs in the brake line you'll have stopping power on at least one side of the car *as well as* in the front and the back. All of these little pieces of mechanical genius went into engineering modern machines, so I'm at a loss to why you appear to believe that graceful degradation is only possible with electronic systems. It's fully possible with mechanical systems as well, and consistently predictable too.

And by that I mean there's a lot of room for judgement in the inspection of mechanical failure modes, not the kind of system that you want to entrust to an industry mostly consisting of people who didn't finish school.

So to summarise around the pedantry: You want redundant electrical fly-by-wire systems in a car if you're going for reliability and safety. Hell you want that in most of industry too.

It depends. Not for brakes, possibly not for steering either (another part with metal linkages that degrade gracefully). Use it for the accelerator, or the cruise-control, or the gearshift (all of which traditionally uses cables). Use it for the power windows, the fuel-filler (or recharge port) cap, the bootlid, the bonnet catch (all of which also uses cables).

But for brakes and steering, there is no point because you have to have the existing system in place anyway, so adding electronics only adds more failure points and does not remove any existing failure points. Brakes and steering are still going to come down to "use this piston/lever/gear to induce motion along this axis".

To be honest, more complexity does not necessarily mean less reliability - a manual gearbox, for example, is a great deal more complex with many more parts than an ICE, yet they tend to outlive ICE's by a factor of two. A lot of the reliability comes from using a design for decades and refining it each generation.

Re:Really?

[stealth_finger]

Yeah I could get behind that. Actually, the last car I got, I could not for the life of me find the control for the side lights so I did what a lot of people do seem incapable of and checked the manual.

Re:Really?

[Alioth]

I know how to change a tyre (and other things, I've done a cylinder head job a couple of times), but the trouble I've had with it, especially at the roadside, is the gorillas at the last tyre place put the wheel nuts on with a ludicrously high torque setting. Last time I needed to do it, I had to use "The Equalizer" (one of those big 1 meter long socket drivers) and pretty much put my entire body weight on the end of it before the lug nut would even budge. All of the nuts were done up like this. One of the studs actually snapped because the overtorquing had weakened and stretched it.

Re:Really?

[bluefoxlucid]

Midbrain is slower than reflex, yes; but it doesn't involve the part of my brain that needs to think and decide. Your entire argument is the prefrontal cortex shuts down and the midbrain goes, "MASH PEDAL HARDER!" The midbrain is capable of complex actions like smoking, scratching your balls, picking at hairs, opening (and finding) bags of food and proceeding to eat, and so forth. When under stress, your midbrain--the part that kicks in and does things when panicking--will start to do these things without alerting you.

This is why martial artists train so damn much. When you lung at someone with a knife, they automatically move forward, avoid the knife, block your arm, punch you in the chin to deflect the rest of your body (it follows the head to avoid neck damage), duck under your arm, put your arm in a lock, and knock you down to your knees. That's not a lot of thinking and planning; that's the brain panicking, assessing the inputs, and applying something it's done thousands of times in similar situations. It might be a long and complex chain of actions requiring the continuous response to sensory input, and the brain can handle that just fine without you being aware of anything more than "oh shit guy is trying to kill me with knife!"

Again: My entire sympathetic nervous system, my hindbrain, and my midbrain have been exposed for 15 years to the various sensory inputs of driving. The 4 inch elevation difference between brakes and gas, the 1 gram accelerator versus the 250 gram brake, the feeling of the car moving forward or backward versus stopping, the visual input of things beyond the windshield, the feeling of torque on the steering wheel when the car moves, all of it. These parts of my brain and even farther down (spinal column...) reflex to the car taking an inappropriate action by making an appropriate response. Granted, I won't automatically grab the handbrake if the foot brake fails; my entire driving experience has primed my system to move onto the footbrake if braking is desired, even if I've mistakenly stomped on the wrong pedal and gotten inappropriate feedback.

Without this, I would constantly crash into things.

Re: Really?

[ceoyoyo]

If you're driving a jeep in Jurassic Park.

Re: Really?

[n0creativity]

I would contend that even 'do nothing' is a dangerous option at times. That would mean all user input is ignored. No braking, no accelerating. This is the inherent problem with non-mechanical systems in applications such as a vehicle. There is NO good option for a 'default failsafe state' without the ability to interpret the context of the situation. Imagine your pulling out onto a busy road, your car malfunctions and goes into the 'do nothing' state. You can't accelerate, you can't brake... there you are coasting across a busy road praying that you don't get demolished.

Re:Really?

[MooseTick]

"I am not saying people shouldn't be allowed to use their car as a deadly weapon"

So you ARE saying people should be allowed to use their car as a deadly weapon

Re: Really?

[bigpat]

Not by default. Through a manual override, sure.

Re:Really?

[drinkypoo]

It's not about being responsible. Humans are fallible.

It is about taking your fallibility into account when making decisions. People who are in the position of having to hit the brake at the last second have already been irresponsible in about 99.99% of situations. They should have looked ahead like a responsible person, seen that the way ahead was congested or complicated like a responsible person, and slowed down ahead of time like a responsible person. Instead, they behaved irresponsibly, and got themselves into a crisis situation they could have avoided. All of this is clearly irresponsible behavior which occurred before their "mistake" of hitting the wrong pedal.

If you're driving while distracted, pull the fuck over until the distraction passes. If my lady is yelling at me while I drive, I don't just keep driving. I pull over. If something falls on the floor of the car and starts rolling around, I don't just keep driving. I pull over. I could go on in this vein but it's boring already.

Now to be fair, I've made my mistakes before, but they were always presaged by some irresponsibility. Driving while tired, driving too close, whatever. They don't just fucking happen. They are made to happen, by irresponsible drivers.

Re:Really?

[thegarbz]

You don't need to drive irresponsibly or have any immediate history of impairment in order to cause an accident. Just as in this case.

I agree there's basically no reason someone should be in a position to rear-end someone. Tailgating impatient arseholes are accidents waiting to happen (and can you even call it an accident when someone drives like a reckless idiot on purpose?). But in this scenario we're talking about a basic human mistake of a foot put down on what was thought to be a break when it's not, and the standard human response, my foot is on the break if I push harder the car will stop!

Those are not accidents that are avoidable with training, or responsibility. There are basic humans mistakes which are just that. Immediate and temporary, not prefaced by recklessness or intent.

Re: Really?

[Immerman]

I will say I've usually only seen that done when someone pulls in front of them going much too slow. They do have far slower breaking and acceleration available than most drivers, and a lot of idiots don't understand that. There's a reason you're supposed to stay far further in front of a semi than a car.

Re: Really?

[delt0r]

but as it's a piece of metal, it's not going to suddenly snap. Metal doesn't do that.

Apparently you don't know shit about metal, metal fatigue, metal corrosion, or metal failure modes. Metal can and does snap. And well for example it doesn't even tell the full story. My breaks totally and instantly failed on an old hillman hunter 1975 car when the *metal* hydraulic line burst.

Re: Really?

[Toshito]

Braking: chose the higher value of the two, log an error if they differ by more than a given margin

That would be the way to go with "brake by wire", but so far cars on the road still have good old hydraulic brakes with vacuum assist, so you can measure the braking force (to activate ABS) but it will always be directly proportionnal to the force applied on the brake pedal.

And I hope that cars will keep hydraulic brakes for the future, I don't ever want a car with brake by wire!

logs

[Anonymous Coward]

Because if its in the data logs thats cast iron it was the meatsack's fault right?
Not even remotely possible say the pot on the accelerator went bad contact, demanded max throttle and that left the audit trail in the logs?

I know we like tesla here, but we've also seen edge cases do the same sort of thing, then someone produces a log that proves something different than what the log does and nobody challenges it and it grows into a worse problem by ignoring it.

Re:logs

[lakeland]

No...

This proves her story is wrong. It does not prove Tesla innocent. There are other scenarios which would have lead to this sensor readout and put Tesla at fault such as a faulty sensor or something jamming the accelerator down.

However we now have on once side someone that is known to lie and on the other side we have Tesla which does not have a past history of lying. So without any further information I'm be inclined to believe the 'meatsack' is at fault.

Re:logs

[Xest]

"Not even remotely possible say the pot on the accelerator went bad contact, demanded max throttle and that left the audit trail in the logs?"

But why would it demand max throttle? we're pretty good at error detection in electronics nowadays, we've been doing it for a long time now. If something is wrong the default will be zero acceleration not maximum acceleration.

Effectively then the only input that can trigger maximum acceleration is depression of the pedal.

This doesn't mean that something couldn't be dropped on the pedal, or that the pedal can't get mechanically stuck, but neither of those things are being claimed here. What's being claimed is that auto-pilot did it, but auto-pilot wasn't engaged and something pressed the pedal.

Re:logs

[dbIII]

Because if its in the data logs thats cast iron it was the meatsack's fault right?

That's what airlines have been saying for decades and now similar technology is in cars.

Re:logs

[bickerdyke]

Because if its in the data logs thats cast iron it was the meatsack's fault right?
Not even remotely possible say the pot on the accelerator went bad contact, demanded max throttle and that left the audit trail in the logs?

I know that every single one who claims to know what happened or did not happen is talking a huge pile of BS.

Ruling out a sensor malfunction is BS, too, but so is ignoring that human error is more likely than a sensor that should be designed to fail safely gets stuck at 100%

Re:logs

[michelcolman]

If something had dropped on the pedal, I'm sure it would have slid right off as the car accelerated. It was a P90D, it accelerates with about 1g.

For it to get mechanically stuck, it would have had to be pushed in all the way first. Not very likely either.

I'd guess it's an 80% chance of her pressing the wrong pedal. 20% chance someone screwed up big time by not putting enough redundancy into the pedal.

Re:logs

[Rei]

Would you explain what lies you're thinking of for the rest of us?

Re:logs

[bluefoxlucid]

It is impossible for a potentiometer to fail in that manner intermittently. The potentiometer would have that failure in that position 100% of the time.

With Experience of Similar Incidents...

[ytene]

I'd be inclined to agree with you but for one thing... A few years ago Tesla let BBC Top Gear test a Roadster, and Jeremy Clarkson lampooned the vehicle in a way that annoyed Elon Musk. Ever since then Tesla have put a *lot* of data capture capability and performance monitoring into all of their vehicles, specifically to stop these sorts of claims.

If Tesla are saying that the telemetry from the black box shows 100% throttle, then at this juncture, I'd be inclined to believe them.

Years ago I spent my spare time helping a friend run his garage business, which included running a contract with a local Police force to recover accident-damaged vehicles. I saw numerous examples of situations in which drivers of automatic cars [and all Teslas are automatic by default] encountered something unexpected on the road. Their first instinct was to slam down on the brake pedal, but you would be amazed at how many managed to hit the throttle by mistake. In the panic and shock of an event, the body can lock up involuntarily, especially, if you think about it, if your car suddenly shot forward under the full acceleration that a Tesla is capable of...

It's way too early to say without more concrete data, but based on the above two points [knowledge of Tesla's extensive telemetry and personal experience of real-world examples like this] my "Occam's Razor" punt would suggest that something happened, the driver panicked, hit the wrong pedal, and the rest is history...

Re:With Experience of Similar Incidents...

[Anonymous Coward]

I did this a few weeks ago.... I was driving along a fairly empty road and had to go past a parked vehicle, so I tapped the brake and to my horror the car went faster! I realised afterwards that I'd been driving for a while and I wasn't sitting completely straight to the wheel, so my feet weren't aligned with the pedals. Then I moved my foot across, too far, and now I was punching between the clutch and the brake (pressing both pedals, but weirdly because they have different pressures) - nothing was working, total madness! It was dark and i couldn't see my feet, and I sailed past the parked car at speed before I finally realigned and got back in control.

Whereupon I straightened myself up, slowed down, and spent the rest of the journey muttering shit...shit...! It'd never happened before, I didn't even consider it a risk. Scary.

Re:With Experience of Similar Incidents...

[Doppler00]

I'm a little suspicious of it being exactly 100% max throttle. If it's an analog A/D you'd think it would be like 100.01% or 99.95% or some in-exact value. It would also be essential to record the intermediate values from 0% to 100%. A jump from 0% immediately to exactly 100% seems a little suspicious to me. Or it could just be the software discretizing the analog data from the pedal at a low sample rate. We don't have the source code to really know.

Re:With Experience of Similar Incidents...

[dbIII]

Their first instinct was to slam down on the brake pedal, but you would be amazed at how many managed to hit the throttle by mistake

In "The book of the Ford", a "missing manual" type book for the Model T Ford written some time around 1920, it mentions that the Model T Ford was designed to stop when both pedals were pressed down at once due to this instinct.
I say both pedals since there was no clutch pedal. It used brakebands instead of a clutch and all gear changing was handled by hand instead of hand and foot.
I've never driven one of the things but I have played with a gearbox from one - epicylic - not so different to a modern automatic transmission as would be expected.

Re:With Experience of Similar Incidents...

[Rei]

I don't know about Tesla's design, but drive-by-wire cars often have, in addition to one or more throttle position sensors, a "closed throttle position sensor" and a "wide open throttle" sensor, each corresponding to precisely 0% and 100% respectively.

Even if Tesla didn't use these, one would expect that the DSP driving the throttle sensor(s) would be calibrated to run from 0 to 100% and truncate any values over that, rather than feeding out-of-spec data onwards into the system.

Furthermore, there is nothing that says that there were no intermediary throttle positions logged.
 

Re: With Experience of Similar Incidents...

[michelcolman]

I imagine "100%" means "the value that corresponds to max pedal pressure" and a short circuit will give something like 150%.

Re:With Experience of Similar Incidents...

[michelcolman]

Come on, do you really think their own actual technical logs showed exactly 100%? They saw some value which may or may not be anywhere near "100" and translated it to a human-understandable "100% throttle" meaning "the pedal was pushed all the way down".

Re:With Experience of Similar Incidents...

[jabuzz]

That all depends on the size of your feet. Plenty of people especially women with small and narrow feet can get their foot between the peddles with ease on lots of models. Just because you have feet the size of Micheal Phelps don't assume everyone else does too.

Re:With Experience of Similar Incidents...

[Rei]

If that's the case, they are logging the wrong thing.

This is a car, not a computer. There are dozens of separate processing units in a modern car. They're not all built into the system that does the logging, as if it could grab whatever data it wants from any of them. Each processing unit is compartmentalized to a specific task. And Tesla, like any other manufacture, does not design all of them.

It's actually rather inefficient from a systems approach, but it's the way things are done in the auto industry, and it's not going to be easy to change that. Each independent unit is made by some OEM or another, who are heavily invested in the current "cluster of independent boxes" approach to car design. And government testing regulations on many components strongly complicate any attempt to combine them with others. Drive-by-wire units (including throttle position sensors), like all safety-critical systems, fall into a very heavily regulated/tested category. And they generally *don't* want unnecessary data transmitted on the network, because everything you send represents a new possible source of problems (unintended data floods, incorrect responses in handling it, etc)

What you're asking for is akin to saying, "My computer should be logging the intensity of the reflected laser light on my mouse and calculating on its own how the mouse is moving". Except far less realistic than even that request.

Re:With Experience of Similar Incidents...

[sociocapitalist]

I'd be inclined to agree with you but for one thing... A few years ago Tesla let BBC Top Gear test a Roadster, and Jeremy Clarkson lampooned the vehicle in a way that annoyed Elon Musk. Ever since then Tesla have put a *lot* of data capture capability and performance monitoring into all of their vehicles, specifically to stop these sorts of claims.

The problem is that the sensors are recording what happens but not why it happens. The sensor can say that the throttle was 100% but it doesn't actually record the movement of a biological leg and foot - it assumes it.

Toyota has recalled cars because of the gas pedal sticking. If that were to happen in a tesla, the sensors would show the throttle going to 100% and would blame the driver when in fact the car was at fault.
http://abcnews.go.com/Blotter/...

Re:With Experience of Similar Incidents...

[michelcolman]

I expect Tesla to start outfitting all new cars with a camera in the footwell.

Re:With Experience of Similar Incidents...

[DrXym]

I'd be inclined to agree with you but for one thing... A few years ago Tesla let BBC Top Gear test a Roadster, and Jeremy Clarkson lampooned the vehicle in a way that annoyed Elon Musk. Ever since then Tesla have put a *lot* of data capture capability and performance monitoring into all of their vehicles, specifically to stop these sorts of claims.

I have a broken console controller that insists I'm pushing on the left stick when I'm not at all. If the console were logging this info then it might claim I initiated this action. Anyone watching me could see that I wasn't.

The same could hold true for a car's accelerator. It is necessary for the logs to have separate channels that correlate user actions to backend commands. e.g. a camera filming the pedal actions is a separate channel, as are resistive pressure pads inside the rubber of the pedals themselves. These channels could be logged with the command and demonstrate a simple cause and effect - user put their foot down (as seen from side channel) and the car sent a command to accelerate. It's an unambiguous way to see what the hell happened.

But if all they're logging is the backend command then it is far harder to prove. I suppose they could construct a graph showing acceleration % rising and falling over time. That might strongly indicate a human or a computer fault - e.g. if it went from 30 to 100% instantaneously, or relatively smoothly as a human might do. But it wouldn't be as unambiguous as a separate channel.

Re:With Experience of Similar Incidents...

[michelcolman]

They did say "abruptly", which is exactly the word one would use if it went straight to 100% from one reading to the next. I don't know at what frequency they record this value, though.

Re:With Experience of Similar Incidents...

[michelcolman]

The Tesla cuts out the accelerator if you press both pedals at the same time, and issues a warning message. So do most other modern cars.

Re:With Experience of Similar Incidents...

[Rei]

They did say "abruptly", which is exactly the word one would use if it went straight to 100% from one reading to the next

It's also exactly the word one would use for a person slamming on the accelerator with their foot.

Nobody says anything about any sort of unnatural transition between non-depressed and fully depressed pedal states; you're writing that into their statements.

Re:With Experience of Similar Incidents...

[simplypeachy]

The mark of a good driver is one who tells themself off for a mistake and learns from it.

Re:With Experience of Similar Incidents...

[Rei]

Not in America, where most cars are automatics.

Re:With Experience of Similar Incidents...

[Rei]

Why are you assuming that this is based on a single sensor?

Re:With Experience of Similar Incidents...

[Dr. Evil]

Good of you to admit it.

I did it once on a rental car after driving for 12 hours... We were on loose gravel and only created a shower of dust. I caught the mistake instantly, but it was enough for my passenger to take over driving. If it were clean asphalt and I were in a sportscar, it would have been a wreck.

FTA: "She knows the difference between brake and accelerator pedal. " - it's amazing how people attribute it to knowledge and discredit "not knowing" as a question of intelligence. The car was 5 days old, it takes more time than that to become intimately familiar with the car.

The reflex reaction to the car lurching forward when you hit the brake is.... hit the brake harder...

And it's ridiculous that the article is interviewing her husband.

Re:With Experience of Similar Incidents...

[drinkypoo]

They did say "abruptly", which is exactly the word one would use if it went straight to 100% from one reading to the next.

Maybe you would, but I certainly wouldn't. I would use instantly, which is a much better fit. So is immediately. Both are wor[l]ds better than abruptly in this case.

Re:With Experience of Similar Incidents...

[drinkypoo]

This is a car, not a computer. There are dozens of separate processing units in a modern car. They're not all built into the system that does the logging, as if it could grab whatever data it wants from any of them. Each processing unit is compartmentalized to a specific task. And Tesla, like any other manufacture, does not design all of them.
It's actually rather inefficient from a systems approach, but it's the way things are done in the auto industry, and it's not going to be easy to change that.

Too bad you don't know how cars work. Each module does its own logging, and the accelerator pedal sensor is connected directly to the PCM. They're not all built into the system that does the logging, the logging is built into all of them. When you want powertrain codes, you have to scan both the PCM and TCM in a typical vehicle, and that's not even counting the ABS which also interacts with the traction control.

Re:With Experience of Similar Incidents...

[drinkypoo]

In fact Automatics are selling better everywhere. They were aspirational in China so they rapidly became a selling point, they are taking over in Europe because they deliver superior mileage, and they are even taking over in exotics because DCTs make any human driver look like a tool.

Re:With Experience of Similar Incidents...

[DrXym]

I didn't say anything about the number of sensors but the number of channels. And given the scant info from Tesla it's reasonable to argue from the position that they don't have a second channel. If you think otherwise, then you can provide a link showing what this second channel might be.

Re:With Experience of Similar Incidents...

[cdrudge]

If that were to happen in a tesla, the sensors would show the throttle going to 100% and would blame the driver when in fact the car was at fault.

You're speculating on the operation of the sensor, if it failed, and what mode that failure would have been.

Re:With Experience of Similar Incidents...

[Rei]

Given that multiple sensors is standard for drive by wire systems, you're the one asserting something out of the ordinary, so the burden falls upon you to prove that Tesla is doing something unusually lax. Heck, is it even legal to have single-sensor drive-by-wire systems? The NHTSA regulates bloody everything about safety-critical parts of cars. Have you ever looked at their regulations? It's insane the level of detail they go into.

Re:With Experience of Similar Incidents...

[Rei]

No, a single sensor wouldn't meet the reqs:

S4.2 In the case of vehicles powered by electric motors, the words throttle and idle refer to the motor speed controller and motor shutdown, respectively.

So I'll substitute that in the following text.

S5.1 There shall be at least two sources of energy capable of returning the motor speed controller to motor shutdown within the time limit specified by S5.3 from any accelerator position or speed whenever the driver removes the opposing actuating force. In the event of failure of one source of energy by a single severance or disconnection, the motor speed controller shall return to motor shutdown within the time limits specified by S5.3, from any accelerator position or speed whenever the driver removes the opposing actuating force.

S5.2 The motor speed controller shall return to motor shutdown from any accelerator position or any speed of which the engine is capable whenever any one component of the accelerator control system is disconnected or severed at a single point. The return to motor shutdown shall occur within the time limit specified by S5.3, measured either from the time of severance or disconnection or from the first removal of the opposing actuating force by the driver.

  The text was clearly written with old-school gas vehicles in mind ("source of energy" and such to pull on mechanical linkages), but they're certainly not going to let drive-by-wire vehicles get by with a more lax reading than traditional vehicles.

Re:With Experience of Similar Incidents...

[rwiggers]

I would consider something similar. You would expect the diver to hit the brake. If the brake stays at 0 and the accelerator goes to 100%, likely user fault. If the brakes are not at 0%, certainly sensor fault.

Re:With Experience of Similar Incidents...

[Rei]

And your examples are.....?

Re:With Experience of Similar Incidents...

[DrXym]

*sigh* I'm not talking about single or multiple sensors. I am talking about independent channels where the cause and effect can be independently verified. Why is this so hard to understand?

Re:With Experience of Similar Incidents...

[Rei]

Would you describe what you're meaning to differentiate "sensor" from "channel", then?

Re:With Experience of Similar Incidents...

[bluefoxlucid]

I brake all the time while the clutch is engaged. As long as I don't go under ~500RPM, the car doesn't stall; dropping some speed by braking while in gear won't stall the engine.

Re:With Experience of Similar Incidents...

[bigpat]

Toyota has recalled cars because of the gas pedal sticking. If that were to happen in a tesla, the sensors would show the throttle going to 100% and would blame the driver when in fact the car was at fault.
http://abcnews.go.com/Blotter/...

Good point. A floor mat issue or mechanical issue with the pedal would trigger the sensor and make it appear that pressure was being applied to the pedal. Basically you would need a camera pointed at the foot to really see what happened under there.

Maybe they should just install cameras under the dash pointing at the controls with the price of camera modules being under $5.

Re:With Experience of Similar Incidents...

[sociocapitalist]

If that were to happen in a tesla, the sensors would show the throttle going to 100% and would blame the driver when in fact the car was at fault.

You're speculating on the operation of the sensor, if it failed, and what mode that failure would have been.

Not at all. I am actually assuming that the sensor is completely accurate.

Re:With Experience of Similar Incidents...

[SpiceWare]

all Teslas are automatic by default

"Manual" and "automatic" are terms used for describing how multi-ratio transmissions change which ratio is in use. Tesla only uses one ratio so there's no way to automatically, nor manually, change it. Transmissions that only use one ratio are known as a "simple".

Re:With Experience of Similar Incidents...

[idji]

Plus, remember Elon's other full time job is send rockets to space and BACK. His guys know quite a lot about telemetry, data and safety. I'd believe a Tesla records data better than any other car.

Re:With Experience of Similar Incidents...

[benro03]

There was a study recently about going on "autopilot" while walking or driving (https://www.psychologytoday.com/blog/mental-mishaps/201404/the-dangers-going-autopilot) and how people avoided obstacles, but didn't truly notice them. Part of the experiment was to bend a branch to head height and then place dollar bills on it. People avoided the branch, but didn't notice the money even when it was waving right in front of them. They even a large sign announcing that a psychological experiment that explained what was going on place in the middle of a path. When asked shortly afterwards about the obstacles, people didn't remember them. They just avoided them.
People get into a car and automatically behave as if they've been driving that car or road regularly, even a new one. They zone out, react to a normal occurrence and because of their unfamiliarity with the vehicle/road do the wrong thing.

Re:With Experience of Similar Incidents...

[dcw3]

The reflex reaction to the car lurching forward when you hit the brake is.... hit the brake harder...

But you don't push it to the floor on the first application. And, you most certainly don't get to 100% that the log is showing. There should have been some intermediate number unless they're logging infrequently...it's not an on/off switch.

Re:With Experience of Similar Incidents...

[dcw3]

I'd be interested to know what cars do this. I'm not aware of any, certainly not any gasoline sports car that one would consider taking to a drag strip, where you typically hold the brake down, and warm up the tires a bit. I know for a fact that it's not true for my Dodge Jeep or BMW.

Re:With Experience of Similar Incidents...

[barc0001]

> The car was 5 days old, it takes more time than that to become intimately familiar with the car.

This is a key piece. I was astonished when I bought a new car how different everything felt for a few weeks. I'd had the old car for 12 years and everything was second nature in it but it takes a while to get used to a new vehicle's layout and handling.

Re:With Experience of Similar Incidents...

[barc0001]

> Their first instinct was to slam down on the brake pedal, but you would be amazed at how many managed to hit the throttle by mistake.

There have been a few proposals to combine the gas and brake pedals into one pedal to combat this problem. Here's one variant:

http://www.nytimes.com/2010/08/04/business/global/04pedal.html?_r=0

Another one I saw several years ago would have the acceleration function tied to the pedal being attached to the control arm on a pivot and by pressing the top of the pedal, the car would accelerate. If you panicked, your foot would slam the whole pedal to the floor and the brakes would activate on full. Interesting ideas but I'm sure there would be loads of resistance to adoption.

Re:With Experience of Similar Incidents...

[Rei]

Not sure what happeend to my response, as I already replied to this :P

Too bad you don't know how cars work

No, too bad you don't know how Teslas work. Teslas actually contain a homebrew Linux box specifically for logging and systems management. When people speak of "the logs" in a Tesla, that is what they're referring to. You download them by plugging in a USB stick. Accelerator stats are are in field 10 (DR1S), offset 12. Value 0 is 0% throttle, value 255 is 100% throttle. It's logged once per second.

I seriously doubt the accelerator itself actually records a constant, internal high-resolution reading of the outputs of each of its sensors. When Tesla wants to investigate a problem, they pull The Logs(TM) as described above. And that's clearly what they did here.

Re:With Experience of Similar Incidents...

[AaronW]

It took me a while to get used to driving my Tesla model S. Sometimes I accidentally hit the edge of the brake pedal when I mean to hit the accelerator since the pedals are a bit close together. The car immediately beeps at me and flashes a warning message on the dash and does the right thing by not accelerating. I think I'd be less likely to accidentally hit the accelerator than the brake, though. I think a lot of people mess up because they use their left foot for the brake and the right for the accelerator. I learned to drive a manual, so the left foot was for the clutch only.

One thing to note is that the Tesla responds instantly. My previous cars there was a noticeable lag between hitting the accelerator and the car actually moving. On my old Prius it sometimes felt like it took a second or more.

Every car is different and with some cars it takes a while to develop the proper muscle memory.

One problem I now have if I have a rental is I always forget to lock the car. I've gotten so used to my car always locking itself when I walk away from it.

Re:With Experience of Similar Incidents...

[AaronW]

I can attest to this. I've done this on several occasions on my model S. If both pedals are hit it beeps and displays a message and does not accelerate.

Re:With Experience of Similar Incidents...

[AaronW]

In fact in many cases automatics do beat manuals in mileage.

Re:With Experience of Similar Incidents...

[thegarbz]

FTA: "She knows the difference between brake and accelerator pedal." - it's amazing how people attribute it to knowledge and discredit "not knowing" as a question of intelligence. The car was 5 days old, it takes more time than that to become intimately familiar with the car.

That's not really relevant in any case. Knowing the difference does not mean you are magically incapable of making mistakes. People do it all the time. I saw an operator with 20 years experience working on a panel that has been the same for 5 years doing what he does 100 times a day, change a value on a control system. In a momentry lapse of reason his head went ... 95 - 5 is 90, and typed in 5 and hit enter. After the valve slammed shut, and the emergency was over, he was presented with the phallic trophy (cockup award) and we started a 5 day long process of restarting an oil refinery.

It happens, and people make mistakes regardless if they've known a car for 5 days, or 5 years.

Re:With Experience of Similar Incidents...

[rubycodez]

you are joking? SpaceX has had multiple catastrophic failures. Musk knows how to lead teams that make shit that fails

Re:With Experience of Similar Incidents...

[michelcolman]

I haven't actually tried it in my Mercedes, but a lot of companies implemented this after the unintended acceleration incidents where people complained that the car continued to accelerate even though they were pressing the brake pedal. To the general public, it seems absurd that you would want to accelerate and brake at the same time. The general public, of course, does not know about doughnuts, handbrake turns, burnouts, left foot braking, etc... They just want the car to be "safe", and the manufacturers are giving them what they want. Same reason why more and more are implementing automatic parking brakes which make handbrake turns impossible. The general public doesn't need handbrake turns.

Meanwhile, I did some more googling for the exact behavior of the Tesla. Apparently, if you first press the gas and then the brakes, it cuts the gas completely. But in other situations it reduces torque without completely cutting it off. There are conflicting reports, though, and I don't have a Tesla yet, so I'll leave it at that.

Re:With Experience of Similar Incidents...

[Cruciform]

It feels really weird getting on a new motorcycle after driving the same one for a long time too.
You go from feeling like you're part of the bike to just sitting on top of one.

Re:With Experience of Similar Incidents...

[DrXym]

I already have. If it's hard to understand, consider that an aircraft's black box. It tells you a plane lost altitude and crashed. It doesn't say that it happened because the pilot collapsed, or someone hit him on the head and threw the switches or the cabin was filled with smoke. Therefore aircraft have cockpit voice recorders and there is an increasing call for them to have cabin cameras. A second channel of information that can be correlated to the first but isn't captured in the same way.

Re:With Experience of Similar Incidents...

[barc0001]

> I think a lot of people mess up because they use their left foot for the brake and the right for the accelerator.

Seriously? Who does that? Jesus Christ, that's a recipe for disaster when the meatbag panics and stiffens their legs. This is yet another in a long series of reasons that I think everyone should be forced to learn on a stick shift, and maybe even drive one daily until it's muscle memory that the left foot is for the clutch or nothing at all.

Re:With Experience of Similar Incidents...

[barc0001]

I can imagine. I like to drive my car a little sporty, and when I got my new one I was very uncomfortable with the feeling of how it reacted differently at speed to turns for a few days. Slightly bigger car and different weight distribution, it just felt... off. I would think that would be worse with a different bike.

Re:With Experience of Similar Incidents...

[EndlessNameless]

Or you could look at the logs indicating 100% accelerator activation and 0% brake activation at the time of the collision.

If the vehicle "drove on its own" without autopilot commands registering anywhere, then either two sensors failed (both the accelerator and brake were wrong) or else the driver accidentally depressed the wrong pedal.

If I have to choose between a freak dual-sensor sensor failure and user error, I'm going with user error. I'll be right over 99% of the time.

Re:With Experience of Similar Incidents...

[avm]

There are a surprising number of people who do this. Oddly enough, even people who are otherwise tolerably decent drivers are guilty. The common point seems to be not learning to drive in a manual transmission vehicle, or not owning one early on in their driving career.

Re:With Experience of Similar Incidents...

[sociocapitalist]

Toyota has recalled cars because of the gas pedal sticking. If that were to happen in a tesla, the sensors would show the throttle going to 100% and would blame the driver when in fact the car was at fault.
http://abcnews.go.com/Blotter/...

Good point. A floor mat issue or mechanical issue with the pedal would trigger the sensor and make it appear that pressure was being applied to the pedal. Basically you would need a camera pointed at the foot to really see what happened under there.

Maybe they should just install cameras under the dash pointing at the controls with the price of camera modules being under $5.

They could even defray the costs by selling the resulting upskirt videos to porn sites.

Re:With Experience of Similar Incidents...

[MondoGordo]

IANAD but "abruptly" is a word that one might use if it went straight to 100% from one reading to the next ... instantaneously is the exact word I would use.

Re:With Experience of Similar Incidents...

[drinkypoo]

No, too bad you don't know how Teslas work. Teslas actually contain a homebrew Linux box specifically for logging and systems management.

The claim was not "Teslas", the claim was "a modern car".

This is a car, not a computer. There are dozens of separate processing units in a modern car. They're not all built into the system that does the logging, as if it could grab whatever data it wants from any of them.

I didn't go back and read the parent comments, because I should not have to. Comments should stand on their own. The Tesla may well work like this, I'm sure it does, but it is unique among modern cars in that case.

Kind of risky

[pablo_max]

I am sure that on the surface it seemed like a good idea to try and blame the car for the couple.
No doubt they wanted to avoid the major increase in insurance premiums and the long wait time for a replacement car.

Nowadays though, with all the tech in cars, it is much easier to check what happened.
Technically, they could now be on the hook for filing a false police report and insurance fraud.

On the other hand, it is possible that it was an honest mistake. It happens regularly that someone "thinks" they have slammed on the brakes, only to have actually smashed the accelerator to the floor. I have even caught the my shoe on the accelerator before on a particular car which the two peddles were very close together.
Another time, I had a cheap ass floor mat that slid up and got stuck in the linkage causing a wide open throttle. Luckily that was a manual shift car.

Are the logs readable by anyone but Tesla?

[FlyHelicopters]

Serious question... is this open information that the driver or owner of the car can read, or is this super secret encoded info that only Tesla has access to?

Do we simply take their word for what the logs say? Is there any way to check via 3rd party that this is in fact what happened and there is a secure means of ensuring the data isn't changed?

This is important, sooner or later it'll end up in court and this will come up. "Trust us" is not an answer.

Re:Are the logs readable by anyone but Tesla?

[Doppler00]

I would guess they don't want people tampering with the black box logs. Although, they could prevent that by just signing the logs from an embedded key as they get written.

Re:Are the logs readable by anyone but Tesla?

[LordHighExecutioner]

Also if logs were readable by everybody, how can one be sure that the autopilot code doesn't contain something like:

if (thisCar->crash()) {
tamperLog->('Autopilot',OFF);
thisCar->autopilot(OFF);
}

Re:Are the logs readable by anyone but Tesla?

[FlyHelicopters]

Open Source... followed by the ability to compile it yourself and compare them...

Re:Are the logs readable by anyone but Tesla?

[FlyHelicopters]

Sure, but I said "readable", not writable :)

All I'm saying is that everyone seems to be given Tesla a level of trust that they would never give Ford or GM.

Re:Are the logs readable by anyone but Tesla?

[ADRA]

Subpoena?

Re:Are the logs readable by anyone but Tesla?

[FlyHelicopters]

Subpoena?

Fair enough, but can you trust that what they hand you is what they really are?

And since you own the car, shouldn't you be able to read your own logs?

Re:Are the logs readable by anyone but Tesla?

[Opportunist]

Well, in all fairness, Tesla never cheated with exhaust parameters...

And they're pretty credible in that area.

Re:Are the logs readable by anyone but Tesla?

[AmiMoJo]

You can read the live throttle position over the OBD-II port, but not the historical data. Might be useful to do some tests to see if the throttle position sensor has failed or is intermittent.

Re:Are the logs readable by anyone but Tesla?

[Anonymous Coward]

You can copy the log to a USB stick yourself ( https://upload.teslamotors.com/ tells you how to do it ) and a quick search on google shows a pdf file with parts of the binary log format decoded ( at least for a 2008 Tesla Roadster, I'm assuming the format doesn't change all that much ).
The pdf notes that Tesla has a 1 Hz logging frequency, so it'll be difficult to differentiate between user error and sensor error.

Re:Are the logs readable by anyone but Tesla?

[Dragonslicer]

This is important, sooner or later it'll end up in court and this will come up. "Trust us" is not an answer.

If it comes up in court, Tesla would have to hand over all of the logs, or possibly even the physical devices if there's a way for others to extract the logs.

As for the "what if they alter the logs" question, that would be more formally called Falsifying Evidence, and it would be the most monumentally stupid thing they could ever possibly do. Judges tend to get just a bit touchy when someone in a trial falsifies evidence.

Re:Are the logs readable by anyone but Tesla?

[crt]

It's a tar file. Instructions for downloading the logs manually here:

https://upload.teslamotors.com/

Re:Are the logs readable by anyone but Tesla?

[F.Ultra]

The car should be available for 3d party examination?

Re:Are the logs readable by anyone but Tesla?

[AaronW]

That's for the Roadster. It doesn't work with the newer cars since the data is accessible wirelessly.

Re:Are the logs readable by anyone but Tesla?

[AaronW]

My Model S has a compliant OBD2 port. The only thing that seems to work on it is power. None of the scan tools I have used work on it. They have a special Ethernet diagnostics connector they can access or they can just access everything wirelessly.

So Tesla tracks everything to do with your car...

[FlyHelicopters]

So Tesla tracks everything to do with your car, people here seem to love them...

Yet bring up Microsoft and Windows 10 and all that tracking and everyone goes all crazy and how "evilz" MS is...

What's up with that?

Solution?

[jsh1972]

Couldn't this be fixed by installing a photocell or proximity sensor or something that would tie in with the test of the data to confirm if it was the pedal depressed or it was a bad potentiometer? Bad pot gives 100% throttle, photocell confirms pedal was/wasn't depressed? It couldn't be that hard and would go a long way towards settling this one way or another.

Re: Solution?

[jsh1972]

*rest of the data... mobile lack of preview strikes again.

Re:Solution?

[Overzeetop]

It would, but you'd be introducing another sensor for the sole purpose of validating a data point which has about a 1 in 100,000+ chance of occurring. With the cost of engineering, software tie in, parts and cabling, you're looking at millions of dollars to validate a very simple mechanical device. If you put one of these in, you need to also put in a brake sensor, and steering wheel sensor, an attentiveness sensor, a... the list goes on forever to create a fully redundant check on every sensor in the car. And then what happens if the sensor breaks? Do you have a sensor for your sensor to ensure its functioning? Do you immobilize the car until the sensor is fixed if there is a fault? How much cost will be associated with sensor failures and maintenance - not to mention poor press if your safety feature immobilizes even a single car (these are 1%ers we are inconveniencing, not some store clerk with a $500, 20 year old Camry that gets fidgety from time to time).

Re: Solution?

[jsh1972]

::shrug:: I'm sure the accountants would draw the line somewhere ;-)

Re:Solution?

[gweihir]

This will not be your average ElCheapo potentiometer as sensor on the pedal. Also, the A/D conversion will be by far fast enough to detect this and you would never use the full range when using a potentiometer as a sensor anyways and hence can detect failures. This is pretty much out as source of the problem.

Log transcript

[Snufu]

Driver: "Autopilot, I need you to stop until the high speed cross traffic has cleared."
Autopilot: "I'm afraid I can't do that, Dave."

Doesn't have to be autopilot

[Anonymous Coward]

If it suddenly went up to 100% straight, then that quite possibly means sensor error. Not autopilot, but not driver error either. We've seen a few cases where some vehicle (including a public transport bus, with not even a driver in the bus at all) "suddenly" started to move. Some of those didn't even have any autopilot at all, just electronic acceleration control. And so a sensor fault is not merely impossible.

In fact, I'd say maybe it's a good idea to limit the range reachable by foot of the sensor to less than 100% (95% or so), so that if it suddenly shoots up to a hundred then that clearly is a sensor fault.

Re:Doesn't have to be autopilot

[michelcolman]

In fact, I'd say maybe it's a good idea to limit the range reachable by foot of the sensor to less than 100% (95% or so), so that if it suddenly shoots up to a hundred then that clearly is a sensor fault.

I would hope it already does that, and "100%" just means "the maximum that can be achieved by pushing in the pedal". Otherwise whoever designed that system should be fired and forbidden from designing anything ever again.

Re:So Tesla tracks everything to do with your car.

[SJ]

Intent counts for a lot.

History has shown us that Microsoft will do everything they can to screw with their customers in any way possibly.

So far, Tesla has been incredibly customer friendly. Until that changes, the data they collect serves to make their products better.

Re:So Tesla tracks everything to do with your car.

[FlyHelicopters]

So far, Tesla has been incredibly customer friendly. Until that changes, the data they collect serves to make their products better.

When that changes...

Tesla is a public company... Google made the "don't be evil" promise once as well, it'll never last...

Re:So Tesla tracks everything to do with your car.

[mrmittens]

You don't generally surf for porn on your car (and if you do, you're a legend/nutter).

Re:So Tesla tracks everything to do with your car.

[Anonymous Coward]

... all that tracking and everyone goes all crazy ...

Windows is designed to provide general-purpose computing. For most people that means digital records of their mail, finances, legal statements, private photos, are stored on a Windows computer. Are people using their in-car computer to handle those documents? When that happens, you will have a valid argument. Tesla knowing what movies people watch and where they go is worrisome but nowhere near the invasiveness possible for Microsoft.

Re:So Tesla tracks everything to do with your car.

[stephanruby]

That was my question exactly.

Why does no one object when I place a camera on the floor of a factory for safety reasons?

But everybody gets hysterical when I place that same camera in the employee toilet looking directly at the employees taking a dump?

What's up with that?

Re:So Tesla tracks everything to do with your car.

[Opportunist]

I MS only gathers the data after a trojan hit me and only after I explicitly tell them to do so, we'll talk.

Heals

[Anonymous Coward]

If it was culturally acceptable for males to wear heals once more, I can guarantee those numbers would be a lot more even.

Re: Heals

[Thruen]

Winner: Most foolish statement of the day. If you are choosing to wear footwear that impedes your ability to drive so you can be fashionable, you need to rethink your priorities before you kill someone just so you can look nice. And if youre suggesting most women put fashion before safety, Id suggest youre at least as sexist as anyone I know.

Re: Heals

[Anonymous Coward]

His suggestion was that *everyone* puts fashion before safety, and that the difference is primarily because social norms prevent males from wearing similar footwear. That you somehow interpreted it as "women are the only ones who put fashion before safety" is intriguing and rather telling.

Re:Heals

[Anonymous Coward]

You can't wear heals; they can only be cast by a cleric.

Re:Heals

[Big Hairy Ian]

Round here it is culturally acceptable for men to wear heals they just find them too damned uncomfortable

Re: Heals

[operagost]

Mythbusters tested the shoe myth, and even though these were MEN wearing women's high-heeled shoes (and we hope they don't wear them often) they had no statistically significant difference in performance between them and any of the other shoes.

Re:Heals

[CrankyFool]

That's racist.

Re: Heals

[Cederic]

I don't drive in my five inch heels. Even though they're boots and a pain to remove, I take them off and drive in bare feet.

It'd be pretty fucking foolish not to.

Re:Heals

[Flea of Pain]

Technically it's classist.

Re:Heals

[kimvette]

This is why if you're wearing stiletto heels, you should keep a pair of sneakers in the car for driving - or, at least drive barefoot (which IS legal, and safer than wearing heels) . Chunky/stacked heels and wedges don't present the same problem.

Yes, 99%+ of the time, you'll manage the pedals just fine, but that one time you don't when the heel catches on the rug, you'll regret not dealing with the inconvenience of changing your shoes when you get behind the wheel.

I don't understand why so many women don't get it. It's a simple concept and adds only seconds to your travels.

And yes, little things like this (disregarding safety in favor of fashion and convenience) does make you a bad driver.

Re: Heals

[kimvette]

Same here. Chunky heels and wedges don't present this problem, but stilettos do. Changing shoes for driving is an inconvenience, sure, but getting into a collision (I refrain from using the word "accident" as most collisions are avoidable) is a bigger inconvenience.

Re: Heals

[kimvette]

But... were they wearing shoes two sizes too small, where the heel has worn and can snag the floormat?

Or, were they sized perfectly, and brand-new, so that they aren't super slippery and don't snag the floormat?

I liked Mythbusters, but some of their methodologies are questionable.

Re: Heals

[Hognoxious]

I can't drive in shoes at all. By shoes I mean formal shoes that you'd wear with a suit. The solution is to keep a pair of sneakers in the car.

I tried driving in bare feet once and it tickled my toes like hell. Had to move the seat forward so I could press with the bit just behind the ball. Maybe if you do it all the time you get used to it?

Re:Heals

[HiThere]

Despite the original comment, it's not just heels that are the problem with shoes, and predominately women's shoes. Sandals are also problematic, as are loafers (though less so).

There is a definite tendency for designs of women's clothing and accessories to be unsafe to use around moving machinery. And to lack conveniences, like pockets. And the excuse, when any is offered, is fashion. Often, however, there isn't even that excuse, and you can't find any other style to buy. My wife is always complaining about the lack of pockets in her pants, and she prefers to buy only pants with pockets...but sometimes there isn't a choice. Or the thing that looks like a pocket turns out to only be decoration when you open the package. Then you get to figure out whether it's worth the hassle of trying to return it...and whether there is anything decent to exchange it for.

It has been said that most designers of womens clothing dislike women. I don't know whether this is true or just a snark.

P.S.: I don't know about currently, but when I was growing up, driving "barefoot" was expensive for a woman, because it was just about guaranteed to put runs in their hose. (I trust the suggestion wasn't that they take off their hose before driving.)

P.P.S.: My wife is so much safer a driver than I am that I have personally revoked my own drivers license. And this is amazing because she doesn't have a good sense of the cars position. But it's still true.

Re: Heals

[nitehawk214]

My mom would just keep some comfortable sandals in the car to switch to.

In the winter when I am wearing waterproof work boots I keep some comfortable shoes in my car to change to. I would rather drive off a cliff than try to drive for an hour with work boots on.

You are damn right when it comes to the socjus people being as sexist as anyone.

Re:Heals

[david_thornley]

How about when your race doesn't have access to the cleric class? In the original D&D rules, player character elves couldn't be clerics, and I'm not sure about hobbits (halflings in the slightly less original rules).

Re:Heals

[kimvette]

Nylons are out of fashion, and have been since the late '90s. Thank god for that... although in winter they were nice. However even if you work in an antiquated workplace asinine enough to have such dress codes, that problem is easily solved by putting on a pair of sneakers.

Re: Heals

[Thruen]

I didn't say only women, I said "most women." The suggestion of "most people" is intended to include "most women," correct? The fact that you somehow interpreted my statement as "women are the only ones who put fashion before safety" is intriguing and rather telling. See how twisting things is stupid and pointless? It twists both ways, every time.

Re: Heals

[Cederic]

I can't move my seat far enough back that using my toes becomes an option.

Re:Heals

[Man Eating Duck]

Driving in high heeled shoes sound so awesomely stupid that it's kind of impressive in its own way. For the sake of the future of humanity I seriously hope that's a very rare cause of accidents indeed.

I arrive at my car wearing stiff MC racing boots every other day, swapping between vehicles at kindergarten. If I actually drove a car wearing them I should lose my driving license, as should anyone doing something similarly stupid behind the wheel. If anyone in a similar situation don't apply one of the obvious solutions they are unfit to conduct a motor vehicle, and should take the bus instead.

Re:Heals

[arglebargle_xiv]

If it was culturally acceptable for males to wear heals once more, I can guarantee those numbers would be a lot more even.

They still wear them today while driving.

Re:Heals

[st0nes]

Driving in high heeled shoes sound so awesomely stupid....

I probably would have hyphenated the compound adjective and pluralised sound, thus: Driving in high-heeled shoes sounds so awesomely stupid...

Re:Incredibly common mistake

[michelcolman]

And according to the article, apparently two thirds are women.

Re:So Tesla tracks everything to do with your car.

[NotInHere]

Am I entrusting my life to my PC? Mostly not. A car can kill me almost instantly.

And the owner?

[Anonymous Coward]

> while his _wife_ was sitting behind the wheel

The gold question is: was the owner sitting behind the wheel too?

Re:So Tesla tracks everything to do with your car.

[Rei]

I know, right? It's like when Miley Cyrus gets naked and licks a sledgehammer they call it "art" and she's an "artist", but when I do it, they're all like "We're going to have to ask you to leave the hardware store" and stuff. I just don't get some people.

Re:So Tesla tracks everything to do with your car.

[FlyHelicopters]

My understanding is that your car is always transmitting all this to Tesla, all the time...

https://forums.teslamotors.com...

This isn't something they "pull" only when you come in for service, they can track and monitor their cars in the world.

I did some Google searching and couldn't find much one way or another...

It's vehicle type ...

[evanh]

The owner lies first then sues second. That's the person type you're gonna get with the Model X.

Re:So Tesla tracks everything to do with your car.

[angel'o'sphere]

The data MS is collecting is transmitted via the Internet to MS servers.
Probably unencrypted.
Probably interceptable by others.

The data Tesla is collecting stays on the black box of the car. until it is needed in case of an accident or for maintanance.

I see a small but significant difference here.

Audi 5000

[Ronin441]

Reminds me of unintended acceleration in Audi 5000s: drivers swore that the vehicle accelerated at full power while they had their foot hard on the brake. Of course, their foot was in fact on the accelerator.

Sudden unintended acceleration#Audi_5000.

Eventually a motoring journalist did the obvious experiment: what happens when you press both pedals at once? At speed or at rest, the brakes won.

Design was a factor: the brake and accelerator were sized and positioned so as to make this mistake easier to make in the Audi 5000 compared to many other cars.

Re:Audi 5000

[Maury Markowitz]

> Reminds me of unintended acceleration in Audi 5000s

I'm shocked I'm this far down the comments before anyone mentioned this.

Re:Audi 5000

[gweihir]

Incidentally, the brakes _must_ be designed so they win in ordinary cars. (Trucks are different.)

Re: Audi 5000

[MasterOfGoingFaster]

Bullshit. My Audi 5000 turbo suddenly accelerated without driver input. It was on cruise control on an interstate highway. I believe a cruise control fault combined with a faulty one way valve on the brake vacuum assist line was the real issue.

Jeremy Clarkson lampooned the vehicle

[NReitzel]

That episode may have annoyed Elon Musk, but it annoyed me too.

The essence of Jeremy's complaint was that the Roadster didn't get close to the advertised range and then made disparaging comments about running out of charge on the way to the Pub.

Except that he was driving the thing on a track at the time, and trying for "best time" laps. Does anyone think that comes close to "normal motorway driving?"

Jeremy, I -hope- you don't drive like that on the way to the pub.

Apologies for a bit off topic.

Re: Jeremy Clarkson lampooned the vehicle

[topham]

Tesla lost because the judge determined the show was entertainment and not a documentary or news program and had no requirement to be factual.

They lied and faked almost the entire thing.

Re:Jeremy Clarkson lampooned the vehicle

[itsdapead]

Except that he was driving the thing on a track at the time, and trying for "best time" laps.

So? This was a Tesla Roadster - a luxury sports car sold largely on the grounds of its performance (which, ISTR, Clarkson did say extremely nice things about). If you are are in the market for a $100k roadster then you might very well plan to drive it to the occasional "track day", spend an afternoon doing power laps, and drive back.

After all, you've already got the Range Rover for shopping, the Bentley for holidays and you just bought little Sebastian a BMW 3 series to get to his lectures... (or, at lest, if you're watching Top Gear, that's your fantasy for 60 minutes). That's certainly the basis on which Top Gear consistently "reviews" cars. They've also been known to point out how many sets of $1000 tyres some supercars vaporised in the course of a track session, how often the car needed filling up while racing the train between London and Cannes, and mocked the compexity of starting the engine on a McLaren, so its not like anything that runs on dead polar bears gets a free pass.

Certainly, taking your Tesla to a track day is a more plausible scenario than wanting to use a Reliant Robin as a re-usable spacecraft (although, hang on, Elon Musk...).

Re: Jeremy Clarkson lampooned the vehicle

[thegarbz]

No, they didn't lie or fake anything. If you go back and read the judgment you see that Top Gear was honest.

They pushed a car claiming it had run out of power into the garage when it still had considerable charge left. They admitted this in court as showing an example of "what could happen" but omitting this on the show.

If you don't consider that lying or faking that says a lot about you as a person.

Re: Jeremy Clarkson lampooned the vehicle

[bentcd]

They pushed a car claiming it had run out of power into the garage when it still had considerable charge left.

As I remember it they had written into the script for the show that they would run the Tesla Roadster around the track until it was out of juice, and then push it into the garage with some suitably disparaging comments. What they found when they got on the track however was that the car had far too big a battery and they far too little time to actually drain it all. This is of course no technical obstacle to filming by the script since you cannot tell by looking at the car that it's still got lots of energy left and so they proceeded as planned. They only got caught because Tesla could tell from the logs it was far from out of battery.

If Top Gear is a consumer information show then this is fantastically dishonest and a big screw you to its viewers. It is not however, it is a sitcom with cars and making jokes is more important than informing the viewer.

The take home of all this is: whatever you see them doing on Top Gear, know that it was scripted and not much influenced by what they actually discovered in the field. If a car wins a "race" it's because the script said it should win. If a car breaks down it's because the script said it should break down. If a car looks like it's somehow better than another car, it's because the script said it should.

Re: Jeremy Clarkson lampooned the vehicle

[jabuzz]

They did lie, they had a shot of a Tesla Roadster being pushed while claiming it had run out of charge. The actual logs from the cars showed that none of the Roadsters made available to Top Gear on that day ever ran completely flat. They got away with it because they are not a documentary or new program and thus have no requirement to be factual.

In the long run it hurt the program as even getting "ordinary" cars for the fastest lap became problematic as manufacturers became hesitant about the "truthfulness" of the program.

Re:Jeremy Clarkson lampooned the vehicle

[ceoyoyo]

No, he drives like that on the way HOME from the pub.

Re: Jeremy Clarkson lampooned the vehicle

[delt0r]

Well you sort of have a point..But this is top gear (the old version) with 3 self professed idiots who have at many times admitted they are in fact mostly ignorant about cars outside of driving em (and lets face it, they are pretty shitty drivers as well). They have never been informative about cars and mostly don't even try to be. Clarson is constantly going on about his god given right to own the biggest gas guzzler SUV and drive like a total asshole. So if viewers think this is an informative show, well then they are the sucker.

Re:So Tesla tracks everything to do with your car.

[wbr1]

Yep. How else are we supposed to make sure our coworkers and employees have not passed out or had a heart attack while defecating. It is all for safety of the user!

Re:So Tesla tracks everything to do with your car.

[ax_42]

Why does no one object when I place a camera on the floor of a factory for safety reasons?

They do object. In most unionised countries, the unions will (in my experience) object as the footage could be used to prove theft, inefficiency etc.

Re:So Tesla tracks everything to do with your car.

[phorm]

Most vehicles log events for a certain time, and may preserve the logs in the event of a crash event, etc. There's a big difference between local logging for post-event analysis and constantly grabbing (TRACKING) your information real-time and storing it offsite.

In TFA, it says "Tesla said it analyzed vehicle logs". Nowhere does it say that those logs were being captured wirelessly, so this is probably the same type of logging that most vehicles do, analysed after the incident.

Hell, most of my servers at home and work "log" tons of data, and that's perfectly OK. What they don't do is ship those logs off to a third-party without my consent or knowledge of what's being sent.

Its actually pretty easy to determine a fault

[burtosis]

Determining if it's a sensor problem is easy in the vast majority of cases with a single sensor. Sure a common sensor failure can register an end point value like 0 or 100%, as those are the most common failures. But most systems can set up where full pedal is only 90% throttle, full off is 10%, and that shows a failure right there. Similarly with good resolution a human in a bumpy vehicle cannot hold a sensor at an exact value for long either, again showing a failure (just like how force sticks in old keyboards rezero).

Simply sampling the pedal 10k times a second is another way. A pedal is a physical device and as such probably cannot be moved through its travel much faster than in 0.1 seconds. You should have one thousand readings showing a smooth transition from unpressed to fully pressed. That is a world of difference from going from unpressed to pressed in 0.0001 seconds - a single sensor reading time sample

Another common sensor fault is getting lots of jitter. Again you can see that the sensor can't be functioning realistically because real pedals cannot move that fast.

Add in a bunch of very simple algorithms and it's pretty easy to approach 100% accuracy in determining if a sensor is feeding correct data or not. It's so trivial and sensor design 101 that I can't imagine all three of these are not already in the tesla.

Re:Its actually pretty easy to determine a fault

[rch7]

You can't imagine that software can have bugs? ;) Really, all software has plenty of bugs and do not always work as intended. And many bugs are random and hardly reproducible.

Re:Incredibly common mistake

[dave420]

Is it because they are women or because women's fashion increases the chance of unsuitable footwear? Let's not jump to some half-baked conclusions...

She got the pedals mixed up

[DirkDaring]

Happens more often than you think. I had a friend that drove like that, he used his right foot for gas, left for brake. Didn't take him long before backing up once and instead of hitting the brake he slammed down the accelerator and smashed into a car. He got that changed real quick after that.

Re:So Tesla tracks everything to do with your car.

[FlyHelicopters]

I wouldn't be so sure...

https://www.teslamotors.com/ab...

Telematics Log Data: To improve our vehicles and services for you, we collect certain telematics data regarding the performance, usage, operation, and condition of your Tesla vehicle, including the following: vehicle identification number, speed information, odometer readings, battery use management information, battery charging history, electrical system functions, software version information, infotainment system data, safetyârelated data (including information regarding the vehicleâ(TM)s SRS systems, brakes, security, eâbrake), and other data to assist in identifying and analyzing the performance of the vehicle. We may collect such information either in person (e.g., during a service appointment) or via remote access.

Remote vehicle analysis: We may be able to dynamically connect to your Tesla vehicle to diagnose and resolve issues with it, and this process may result in access to personal settings in the vehicle (such as contacts, browsing history, navigation history, and radio listening history). This dynamic connection also enables us to view the current location of your vehicle, but such access is restricted to a limited number of personnel within Tesla.

Simple answer

[aepervius]

The car do not phone home to transmit that data to Tesla. It is only saved locally in the car and looked at only in case of accident.

Re:Simple answer

[FlyHelicopters]

I wouldn't be so sure...

https://www.teslamotors.com/ab...

Telematics Log Data: To improve our vehicles and services for you, we collect certain telematics data regarding the performance, usage, operation, and condition of your Tesla vehicle, including the following: vehicle identification number, speed information, odometer readings, battery use management information, battery charging history, electrical system functions, software version information, infotainment system data, safetyârelated data (including information regarding the vehicleâ(TM)s SRS systems, brakes, security, eâbrake), and other data to assist in identifying and analyzing the performance of the vehicle. We may collect such information either in person (e.g., during a service appointment) or via remote access.

Remote vehicle analysis: We may be able to dynamically connect to your Tesla vehicle to diagnose and resolve issues with it, and this process may result in access to personal settings in the vehicle (such as contacts, browsing history, navigation history, and radio listening history). This dynamic connection also enables us to view the current location of your vehicle, but such access is restricted to a limited number of personnel within Tesla.

Re:So Tesla tracks everything to do with your car.

[johnw]

The two are (potentially) very different.

I have no first hand experience of Teslas, but from the descriptions I've seen it sounds like this is a black box type of recording. It doesn't appeared to be used for anything until you crash the car, when it can be used to investigate what happened.

The tracking in Windows 10 on the other hand serves no such purpose. Call Microsoft and say, "I've corrupted my vital spreadsheet. Please check your tracking data and tell me what I did wrong." and you're not going to get much in the way of useful assistance. The Windows 10 stuff exists purely for the purpose of Microsoft extracting value from their customers.

I may be wrong - it may be that Tesla track where you drive and when and then use it to try to sell you stuff, or to sell information about you to third parties. I've seen no evidence of it so far though.

Re:So Tesla tracks everything to do with your car.

[johnw]

In what way has history shown us that Microsoft will screw its customers?

You must be new here.

Re:So Tesla tracks everything to do with your car.

[FlyHelicopters]

I wouldn't be so sure...

https://www.teslamotors.com/ab...

Telematics Log Data: To improve our vehicles and services for you, we collect certain telematics data regarding the performance, usage, operation, and condition of your Tesla vehicle, including the following: vehicle identification number, speed information, odometer readings, battery use management information, battery charging history, electrical system functions, software version information, infotainment system data, safetyârelated data (including information regarding the vehicleâ(TM)s SRS systems, brakes, security, eâbrake), and other data to assist in identifying and analyzing the performance of the vehicle. We may collect such information either in person (e.g., during a service appointment) or via remote access.

Remote vehicle analysis: We may be able to dynamically connect to your Tesla vehicle to diagnose and resolve issues with it, and this process may result in access to personal settings in the vehicle (such as contacts, browsing history, navigation history, and radio listening history). This dynamic connection also enables us to view the current location of your vehicle, but such access is restricted to a limited number of personnel within Tesla.

Re:So Tesla tracks everything to do with your car.

[FlyHelicopters]

I wouldn't be so sure...

https://www.teslamotors.com/ab...

Telematics Log Data: To improve our vehicles and services for you, we collect certain telematics data regarding the performance, usage, operation, and condition of your Tesla vehicle, including the following: vehicle identification number, speed information, odometer readings, battery use management information, battery charging history, electrical system functions, software version information, infotainment system data, safetyârelated data (including information regarding the vehicleâ(TM)s SRS systems, brakes, security, eâbrake), and other data to assist in identifying and analyzing the performance of the vehicle. We may collect such information either in person (e.g., during a service appointment) or via remote access.

Remote vehicle analysis: We may be able to dynamically connect to your Tesla vehicle to diagnose and resolve issues with it, and this process may result in access to personal settings in the vehicle (such as contacts, browsing history, navigation history, and radio listening history). This dynamic connection also enables us to view the current location of your vehicle, but such access is restricted to a limited number of personnel within Tesla.

Log for safer driving

[Applehu Akbar]

Tesla uses logging to check the performance of its automation, but why not have a data log of all driver action and readable car data? This could be implemented flight recorder-style, capturing the last hour or so of data, so that privacy would not be compromised. It could do a world of good in adjudicating accidents.

Accelerator failsafe?

[Cutriss]

Probably won't get noticed 150+ comments deep, but...

Perhaps the default configuration for the pedals should be a failsafe mode where the car is always "under control". When you slam the brake, you trend toward 0 MPH. If you slam on the gas, maybe the pedal interprets 100% as 0%, and applies no throttle. If you're accelerating, you should always have control of the accelerator. Flooring it isn't going to give you much more than 95% throttle would, and you could have a tactile bump at the end of the accelerator play that is easy sensed when you feather your foot, but also easily bypassed if you slam the pedal.

Basically, allow people to still gun it, just not outright "drag racing", and prevent unintended acceleration.

Re:Accelerator failsafe?

[gweihir]

That is not a good idea: The typical situation where you floor the pedal is one where getting all the power available _now_ saves your life: Overtaking with oncoming traffic and you misjudged. Sure, you could prevent that when going at low speed, but once you start to mess with this, you will be facing lawsuits right and left, because some people will find situations where the automatics amplified damage, even if on average they reduced it.

Re:Accelerator failsafe?

[thegarbz]

Perhaps the default configuration for the pedals should be a failsafe mode

As mentioned in a few other places, a car accellerator has no fail safe. The "safe" position is situation dependent, e.g. the safe position changes every time you enter an intersection for a period.

Assuming the pedal failed (which I don't think it did) the way around this is to use 3 sensors and do a mid-out-of-3 control on the result.

" When will people stop lying about..." - Really?

[Assmasher]

People will never stop lying about things they think they can get away with - especially to avoid wounding their pride.
One of the few positives to the surveillance state, the constant tracking of data - is occasionally the ability to point out just like Tesla - "Uh, you're a lying sack of shit..."

Re:So Tesla tracks everything to do with your car.

[The MAZZTer]

Well I guess the question is, can Tesla access these logs at any time? It would seem to me the only reason they accessed and likely had to the capability to access these logs was after the fact to diagnose the crash. Which is fine!

Re:So Tesla tracks everything to do with your car.

[phorm]

Ugh. I really do hope that is optional, or at least tunable.

Re:So Tesla tracks everything to do with your car.

[FlyHelicopters]

I don't know, I don't own one...

I'm simply pointing out that they have the ability to do it... I think this is something that should be adjustable by the end customer...

Potential for money attracts fraud

[sjbe]

When will people stop lying about Tesla's Autopilot mode crashing their cars?

Only when it becomes clear that there is no money to be made by doing so. People lied about Audi's "uncontrollably accelerating" when in fact it was people standing on the accelerator pedal when they thought they were on the brake. People lied about Toyota Prius's too because they though there was a chance for settlement money. I have no doubt Telsa and every other company that comes out with autopilot technology is going to get hit with fraudulent claims as well.

Re:Potential for money attracts fraud

[gweihir]

And hence the event recorder and they will make _very_ sure it records accurately. Eventually all cars will have one and that will curb this dishonorable and greedy behavior.

Re:Potential for money attracts fraud

[Enigma2175]

When will people stop lying about Tesla's Autopilot mode crashing their cars?

Only when it becomes clear that there is no money to be made by doing so. People lied about Audi's "uncontrollably accelerating" when in fact it was people standing on the accelerator pedal when they thought they were on the brake. People lied about Toyota Prius's too because they though there was a chance for settlement money. I have no doubt Telsa and every other company that comes out with autopilot technology is going to get hit with fraudulent claims as well.

The Toyota problem was a real issue, and Toyota actively destroyed evidence and attempted to cover up the problem. From this page:

In April 2013, Betsy Benjaminson, a freelance translator working for Toyota to translate internal documents, released a personal statement about Toyota covering up facts about the sudden unintended acceleration problem. Benjaminson stated she “read many descriptions by executives and managers of how they had hoodwinked regulators, courts, and even congress, by withholding, omitting, or misstating facts.”

Re:So Tesla tracks everything to do with your car.

[angel'o'sphere]

"Countries" are not "unionised".

Industries or "industrial sector" are.

And no: it is pretty common to have camaras installed in saftety relevant areas. There are easier ways to catch theft anyway.

Re:Incredibly common mistake

[Megol]

Especially as there are a lot of studies that indicates women in general are better drivers than men...

Re:So Tesla tracks everything to do with your car.

[houghi]

Why does no one object when I place a camera on the floor of a factory for safety reasons?

Depends where the camera points to, for example. Because there are places and countries where they very much will object if you e.g. point it to where people are working. You better have a damn good reason that you need to monitor that place and a damn good reason why you can't do that without filming your employees.

Re:_wife_

[Megol]

Nope. The only way to interpret your post ATM is indicating that you are a prejudiced idiot and I hope some extra words can put your post in a different light.

The bane of the event recorder

[gweihir]

It will take a while, but eventually people will need to learn that with an event recorder in the vehicle, they have to actually fess up to the stupid things they did.

Of course, honest people did that l along, bit they seem to be increasingly hard to come by these days.

Re:When Will We Get to Audit the Code?

[gweihir]

That would likely put them in jail for an extended period of time...

Re:" When will people stop lying about..." - Reall

[gweihir]

Not all people are like that, and there will be some people that are honestly confused about what happened. But I agree, there are far too many liars around.

0 to 60 in 3.8 sec

[funkymonkjay]

to be expected. what would have been just a roar of the engine will now land you across the street.

Re:Toyota was also blaming drivers.

[gweihir]

Independent code audits. Tesla will make very sure the logs are accurate and that they can prove it. And no, logs, done right, show what was done to the pedal as well.

this report is not credible

[Thud457]

it wasn't even a farmers' market that got ran into.

AEBS, haven't Tesla heard of it?

[Computershack]

This is not good enough for Tesla given the computing ability and the sensors it has in place. My 18 wheeler semi, a 44 tonne DAF CF Euro 6 my company bought along with 100 others in October 2014, is fitted with AEBS. AEBS is now mandatory in all heavy trucks in the EU. The collision the Tesla had is impossible in my truck or any new truck recently sold within the EU unless there is a fault. If the AEBS system detects a collision it automatically applies the brakes and disables the accelerator pedal function. If a truck that is over 18 months old has the ability to prevent this regardless of driver input then why doesn't a Telsa which is lauding its autopilot ability regardless of whether it was the driver or the autopilot in control? Seriously this is "old" tech, the problem has been solved and even implemented in law in the EU so why haven't Tesla implemented it?

Re:So Tesla tracks everything to do with your car.

[guruevi]

All modern cars track everything about your car. The values are shoved in a little box and are readable by the manufacturer. They aren't sent off remotely (yet) for advertising purposes without your knowledge.

Re: Somebody's getting a beating tonight

[eriklou]

Pretty sure there are sensors on the throttle body that confirm the petal position too.

Re:So Tesla tracks everything to do with your car.

[nickersonm]

I believe you can opt out of the program at purchase.

Re:Incredibly common mistake

[Cederic]

Really? Evidence?

My understanding is that men are more likely to die while driving, but women average more accidents per mile driven.

It also depends on your definition of 'better drivers'. I don't see many women driving at Le Mans or competing in Formula 1.

Re:Incredibly common mistake

[Enigma2175]

Especially as there are a lot of studies that indicates women in general are better drivers than men...

No, they show that women get into fewer accidents than men, not that they are better drivers. There are many possible causes for this result. I think there are probably several factors that contribute. For example, in many families the male is the default driver when the family is traveling and is almost always the driver in inclement conditions.

Women do seem to take fewer risks and are, in general, less aggressive drivers which will certainly lower their accident rate, but I don't think that is the only cause.

Re: Somebody's getting a beating tonight

[gnu-sucks]

If it's like my Honda Civic, "drive by wire," then there is a simple sealed variable resistor installed at the pedal lever. If one wire fails, the encoding ADC will either swing to the top or the bottom of the range. There really isn't a "throttle body" in an electric car.

So it is still possible that the throttle encoding circuit/wires failed and the computer logged a "full press" even though it wasn't pressed at all.

If it actually logged 512 or whatever the max is, I wonder if a normal press of the pedal could have achieved full scale? Maybe.

Toyota had plenty of settlement seekers

[sjbe]

The Toyota problem was a real issue

Some of it was real but much was fabricated by would be con artists and ambulance chasers. There were a lot of people making false and/or unsubstantiated claims in the hopes of settlement money.

Re: Somebody's getting a beating tonight

[AaronW]

In most if not all cars there are redundant sensors for the throttle position and the ECU is designed to detect a failure.

Too many drivers

[peterofoz]

This can also occur if there are more than one person in the drivers seat. Pants optional.

Re: Somebody's getting a beating tonight

[Nethemas the Great]

Pretty sure there's no such thing as a "throttle body" on an 100% electric vehicle.

Re: Somebody's getting a beating tonight

[Hognoxious]

confirm the petal position

By default they're attached to flowers.

"abruptly increased to 100%"

[rch7]

"abruptly increased to 100%" - sounds like BS. How on Earth you can now if a pedal was pressed 100%, or one of 2 potentiometers failed and software failed to apply minimum throttle instead of maximum, or it was just one of zillion of software bugs that registered 100% press when there was none? Tesla should release raw data if it has any meaningful & detailed data, then we can see how it changed over time and if Tesla story is credible, was there gradual press on pedal like pressing on brake, or was it some sudden jump to 100 over 5 milliseconds that doesn't look like human input. Right now, it looks like Tesla is covering it up or not taking it seriously.

The middle pedal is not the casseur

[Hognoxious]

I am talking about slowing down, not about breaking.

If you don't slow down in time you might well break something.

Re:So Tesla tracks everything to do with your car.

[AaronW]

All data is encrypted over an OpenVPN connection. I know, I looked at all the traffic my Model S was sending and scanned for any open ports (there were none).

Re:So Tesla tracks everything to do with your car.

[AaronW]

You can turn this off. By default it is off.

Re: Somebody's getting a beating tonight

[gnu-sucks]

Detecting the failure should be easy, I agree. Could be done with simple analog circuitry or in the digital domain.

I can tell you that in my car there exactly three wires going to the accelerator pedal. And since that is the only place I am inputting the desired accel, this is where a redundant sensor would need to go. I'm sure on the gas engine side there are sensors on the throttle air/gas mixture block, so this would show what the engine actually did. And I would imagine the state for the cruise control is recorded too. But from a "person's foot command" point of view, only one sensor.

Re: Somebody's getting a beating tonight

[Cramer]

It's actually the complete opposite. VERY FEW cars have redundant throttle position sensors.

What should be checked here is the slope of the throttle application. Did someone actually push a peddle or did it "instantly" go from a reading of 0% to 100%.

My Log Says You Owe Me 1,000,000

[BrendaEM]

Obviously, the log could not be tampered with, and there only objective people working with it?

Re: Somebody's getting a beating tonight

[nachtelfjeiu]

I'm guessing that's exactly what Tesla checked. Besides I'm much more inclined to believe Tesla. Integrity and honesty are simply virtues of the past for many civilians of a sue and blame culture. Tesla has more to lose by lying.

Classic Straw-Man

[GrantRobertson]

Sounds as if Tesla set up a perfect Straw-man fallacy (and the owner helped): "Because the autopilot didn't do it - it MUST have been the driver." As lots of other posts here show, there are lots of other things that could have gone wrong which could have filled the logs.

That kind of denialism, though the management may think it shields them from bad press, will eventually hurt them far worse than performing a full, transparent investigation.

Re: Somebody's getting a beating tonight

[AaronW]

You would be wrong. Early cars lacked the redundancy but today most if not all have redundant sensors. I know when the whole Toyota fiasco flared up their cars had redundant potentiometers.

http://jalopnik.com/how-electr...
http://www.diycardoctor.com/ac...

With dual potentiometers it is configured such that they behave in the opposite manner such as when the voltage on one increases the voltage from the other decreases and this is checked by the ECU.

It's obvious the people are lying

[Agent0013]

They are saying the vehicle drove over 39 feet of planters to hit the building. The picture shows the plants and grass to be less than the length of the model x. The model x is only 198.3 long which comes out to 16 and 1/2 feet, less than half the distance they claimed it drove. I guess they don't care about the truth or verifiable facts.

Re: Somebody's getting a beating tonight

[samwichse]

That sounds pretty unsafe, given the unreliability of variable resistors.

IIRC, the Civic uses Hall-effect sensors for its TPS, with a pair of redundant ones on the gas pedal.