[jakemoshenko]

Why do we need an accelerator when we know how fast the wheels are turning directly? The first derivative of wheel speed is almost always car acceleration. Is this to handle some edge cases like not turning the brake lights on when we're actually sliding on ice with the wheels locked? Seems unnecessary.

[willis936]

Why is it unnecessary? Isn't the primary goal of brake lights to inform people of deceleration and the secondary goal to indicate when a vehicle is locked stationary?

Showing a speeding vehicle behind you that you've lost control and should slow down is well within the conops of brake lights.

[NikolaNovak]

I think there's an agreement on the goal.

I believe the poster is asking why do we need another sensor to achieve that goal? We already have a speedometer based on wheel motion. A trivial computation gives acceleration.

[hef19898]

Because tires can block, not all tire turn at the same speed all the time, because redundancy is a good thing... There a lot of reasons. And if the number of sensors in your car bother you, well, the early 80s, with carbs and without ABS, are last model years you can buy.

[NikolaNovak]

No need to get snarky :)

Redundancy is a good thing. Do we feel manufacturers are cross-checking accelerometer with the speedometer? If not, there is no redundancy gained.

On the other hand, if your speedometer is faulty, or not working, you'll know pretty fast. If your dedicated accelerometer is faulty, you might have no idea what your brake lights are doing in the back.

Dunno; it just feels "we need to know acceleration, let's add an accelerometer" is a non-imaginative, add-cost, add-complexity idea. Again, if we think ABS and speedometer and the new accelerometer are being cross-checked and sanity-checked, awesome, but I'm just a bit cynical of that, compared to adding the 3rd thing to do the same thing.

[Sohcahtoa82]

> add-complexity idea

You're on a forum where a lot of people think Kubernetes is a good idea.

;-)

[hef19898]

No idea what Kubernetes are, and I am serious here, since I am no software dev. On the hardware side of things, there is redundancy and complexity. Those two might look similar on the surface, they are totally different beasts so.

[vdqtp3]

Many late 80s/very early 90s fuel injection vehicles have shockingly low numbers of electronic sensors also. MAP, O2, TPS, CPS is all you really need. Some of them are even analog.

[tlb]

An accelerometer will show braking when going at constant speed down a hill. Change in wheel speed is better.

[mcny]

I’d say you are braking somehow if you are going at a constant speed downhill, are you not?

[willis936]

What do you save by not having that sensor? Nothing. What do you lose by not having that sensor? The ability for the system to perform its requirements.

I'm assuming you read the context of loss of traction being a requirement.

[oceanplexian]

> What do you save by not having that sensor?

A $9000 repair bill in 6 years when your mechanic tells you “Sorry your car is immobilized and won’t start, the deceleration brake light sensor went and we need to remove the motor and 3/4 of the wiring harness to replace it”. And for anyone who thinks I’m being sarcastic, try owning a BMW or an Audi and you’ll know it’s the truth.

[willis936]

A MEMS accelerometer is pennies. If it is critical then add redundancy and don't accept single fault tolerance. If it is optional or the signal can be estimated from other sources with acceptable error then fail gracefully. This is honestly simple stuff. Accepting less from manufacturers is a bad deal.

[NikolaNovak]

>>A MEMS accelerometer is pennies.

That's an dishonest argument (perhaps not intentionally). NOTHING is "pennies" to a consumer when it comes to repairing a vehicle. More frequently, you pay $1,500 in labour and parts to replace something that costs "pennies" in some bulk manufacturer wholesale catalogue.

Modern cars are getting more and more awesome, in terms of safety and convenience; but the sticker shock when going to dealership for repair is also becoming bigger and bigger, and it absolutely is intertwined with the significant effort to make 3rd party or even self-repair difficult to impossible.

So again, IFF this is an easily replaceable part that is thought-through and cross-checked intelligently with other existing sensors, brilliant. But can you at any level understand my skepticism that any of these are true? :)

[mulmen]

I have owned a troublesome 2010 BMW for 8 years now. No individual repair has been over $2,700.00. Most are closer to $1,500. I was quoted (by a dealer) $15,000.00 for engine repair once but it turned out to be a spark plug. $9,000.00 sounds like a dealer quote. Find a good independent shop.

[Turskarama]

The problem there is owning an Audi or BMW, not the sensor itself.

[WorldMaker]

Relatedly, having a dedicated accelerometer sensor is a baseline dependency of many of other modern features like cross-comparing compass headings and GPS for map heading information (much less any of the Level 2+ "self-driving" mechanics). Most cars likely want one, anyway, even in base models. It's one of the cheapest sensors in a suite of increasingly standard sensors (in almost any form factor of device, not just cars, but phones/watches/toasters/etc).

[chandlerswift]

One more possible reason might be when driving down a hill: When descending a sufficiently steep incline, I may be braking just to maintain my speed! That said, I'm still _feeling_ negative acceleration to maintain a constant speed. (That said, if I'm doing this for more than a handful of seconds, I'll generally downshift, and _that_ doesn't trigger brake lights, despite also causing increased negative acceleration. So maybe regenerative braking is analagous, and fine without brake lights?)

[tetromino_]

> Why do we need an accelerator when we know how fast the wheels are turning directly?

Because the car doesn't precisely know the wheel's radius. The radius depends on the rim, tire, and tire pressure - and the car's operator may have accidentally or deliberately chosen something unexpected for any one of those 3 parameters.

[bentcorner]

You're not wrong but how much does not precisely knowing the wheel's radius change things?

The operator would also experience an incorrect speedometer. For the purposes of a brake light it'd be off (either lighting up too soon or too late) by some amount but I imagine it'd be "close enough" except for the most pathological cases of tire sizes.

[oh_sigh]

You don't need to know the wheel radius, you just need to the rate of change of the wheels RPM. Then the brake light could be set to some conservative value where regardless of any realistic tire size/inflation scenario the brakes will activate at a reasonable deceleration.

In any case, with standard gas/brake pedals on ICEs, I can barely tap the brakes and have my brake lights turn on without any deceleration on my part, or I can do that and still be pressing the gas and so actually be accelerating with my brake lights on.

[mrguyorama]

This is not a problem. We know this isn't a problem, because ABS almost always uses wheel speed sensors to do it's thing, and ESC usually taps into the same wheel speed sensors. Hell, those sensors are often used to tell you when your tires are low on pressure, and yet that doesn't prevent any of the other functions from working, or even affect your speedometer enough to care.

Changing wheel size enough to affect readings of speed does not change the broad slope of the derivative of wheel speed.

[mikestew]

ABS relies on the fact that the computer knows how fast the wheels are turning in relation to each other. Wouldn't seem to be much of a leap to leverage that system to determine if the car is slowing down.

[yjftsjthsd-h]

If the car doesn't know the size and speed of the tires it also can't accurately display the current speed, which seems like at least as pressing a problem.

Edit: Or is detecting deceleration more sensitive? I guess I don't know how precise it needs to be relative to how precise a speedometer needs to be.

[nonfamous]

I don’t think it’s such an edge case at all. I imagine it would be disconcerting to the driver behind you if you slammed on the brakes on a wet road, locked the wheels into a skid, and the brake lights went out. (Shouldn’t happen with modern braking systems, but still…)

[ilyt]

Could be as simple as "the controller doesn't have wheel speed on its can bus"

[Turskarama]

>The first derivative of wheel speed is *almost* always car acceleration.

There you go, you answered your own question. When it comes to safety features, "almost" always working isn't really good enough.