Safety Score

[EVNow]

90K in the USA would be many multiples of that 8% I would think.

Nearly a million cars now sold in US ! @10% - it comes to 100k.

BTW, I don't remember Deepak saying 8% - but we know from a leak database, it was 35% at the end on Q3 '17.

 

[dckiwi]

I think you're good on the 20-40k number but I think the number of 100's is probably lower than 10%.

I hope you're right, because then they really should get through the 100s in the first day or two.

 

[boonedocks]

I still believe that there may be at least one instance where AP may not be used 'appropriately' as EM cryptically said, and therefore not masking an event. In particular, if you override AP with your accelerator, that may turn off the masking. I had gotten used to 'helping' AP start moving, or preventing a hard brake by pressing lightly on the pedal. This may be what is causing some people to have following and FCW events while on AP.

I now consciously avoid touching the accelerator when on AP.

I found this out the first (Monday) daily commute being scored Tesla. Killed my Following Distance horribly that day. Stopped pressing the go pedal while on AP and almost consistent 100's since....

 

[HighZ]

Here is the current top 25 in order of Weighted Miles Driven. We do not have any reason to believe that Tesla will do this, and in fact, Elon said it would go in order of the raw total score, but it is interesting to see that there are not that many 100's in this list.

Also, kudos to the Model X that has driven over 3100 miles with a 97. Granted, I am sure there were a lot of freeway AP miles, but still, there had to be a lot of charging stops.

 

[TIppy]

Oh yes - that’s possible. I briefly forgot they might be using the non-gravity corrected accelerometer instead of finding deceleration using actual speed. Good explanation by @mongo in the below linked post.



BTW, since we don’t really know what method Tesla is using, safe thing to do is to be careful both uphill and downhill.

If they were using the y-axis accelerometer to measure acceleration along the longitudinal axis of the car, then regen would be reacting to that too. Any component of the acceleration due to gravity due to going downhill would actually decrease your effective regen. If you were on a 20 percent downhill grade the accelerometer would indicate you were slowing at 0.2g (the limit of regen), but instead your speed would be constant. So the downhill component wouldn't increase your measured deceleration. To test this, one could regen down a steep hill and record a video of the speedometer to see what the actual change in speed wrt time is. The safety score documentation gives actual change in speed(mph) per second as an example.

I really don't think they use the y-axis accelerometer.

 

[EVNow]

If they were using the y-axis accelerometer to measure acceleration along the longitudinal axis of the car, then regen would be reacting to that too. Any component of the accelertion due to gravity due to going downhill would actually decrease your effective regen. If you were on a 20 percent downhill grade the accelerometer would indicate you were slowing at 0.2g (the limit of regen), but instead your speed would be constant. To test this, one could regen down a steep hill and record a video of the speedometer to see what the actual change in speed wrt time is. The safety score documentation gives actual change in speed(mph) per second as an example.

Its not clear what you are saying here - are you saying they use they don't use the accelerometer but actually use speed to calculate ?

And yes, going downhill on some of the hills for me usually maintains the speed with just regen.

 

[boonedocks]

Here is the current top 25 in order of Weighted Miles Driven. We do not have any reason to believe that Tesla will do this, and in fact, Elon said it would go in order of the raw total score, but it is interesting to see that there are not that many 100's in this list.

Also, kudos to the Model X that has driven over 3100 miles with a 97. Granted, I am sure there were a lot of freeway AP miles, but still, there had to be a lot of charging stops.

View attachment 718399

with a 99 and now over 1,300 miles I may actually have a chance LMAO

 

[tivoboy]

Nearly a million cars now sold in US ! @10% - it comes to 100k.

BTW, I don't remember Deepak saying 8% - but we know from a leak database, it was 35% at the end on Q3 '17.

Well if it’s 1M cars in the USA with FSD capable hardware, then 100K sounds about right. Certainly hard to get a true fleet average, but I’ve seen surveys of 10K plus owners that have indicated anywhere from 10% to 35% within any given purchase quarter when the surveys have been done. Certainly ramped up with M3 sales volume dwarfing model S and X sales, but doesn’t give an easy to relate USA fleet average.

Understanding The Decline In Tesla’s FSD Take Rate - Tesla Motors Club

On August 31st, 2021, “Troy Teslike” tweeted data from his long running survey of thousands of Tesla vehicle buyers, showing a low take rate for FSD, Tesla’s “Full Self Driving” software add-on offered since late 2016. For Q2 2021, that survey shows a worldwide FSD take rate of only 11% across...

teslamotorsclub.com

 

[dckiwi]

Also, kudos to the Model X that has driven over 3100 miles with a 97. Granted, I am sure there were a lot of freeway AP miles, but still, there had to be a lot of charging stops.

Also - I have no idea if this is just me - but my Model X has TONS of FCWs and my Model 3 has none.

 

[TIppy]

Its not clear what you are saying here - are you saying they use they don't use the accelerometer but actually use speed to calculate ?

And yes, going downhill on some of the hills for me usually maintains the speed with just regen.

I'm saying they use an inertial measurement unit for vehicle dynamics. I was pointing out that if they just used a simple accelerometer aligned with the direction of travel, the actual deceleration of the car under regen would decrease going downhill, because the gravity component would be added to the regen and the sum would be limited to 0.2g.

 

[yurmix]

I hope I’m not repeating the question,

Does anyone know if the weight of a bad drive is by the single drive miles or the entire day? Just wondering if I should cancel my plans
I was at at 99 up until yesterday, but todays score is 49 .

 

[jebinc]

I hope I’m not repeating the question,

Does anyone know if the weight of a bad drive is by the single drive miles or the entire day? Just wondering if I should cancel my plans
I was at at 99 today and my morning routine started with a day rated at 49

Drive a lot today! Once tomorrow comes, you're screwed if you don't get today's score up. BOHICA, as they say.... @2021plaid

 

[PDX-Y]

I really don't think they use the y-axis accelerometer.

There are some 30% grades in San Francisco (though not, heh, Tampa, where your profile claims to be). Seems like it should be a straightforward thing for some enterprising geek to go out and test.

But I agree, there's no need for accelerometer input here. Just reading instantaneous speedometer output is more than sufficient. Turning rate gets harder though, depending on how much precision they can read the wheel position. There, I might believe that e.g. "turning hard on a banked track" would count for less lateral acceleration. But then... it probably should, if it's measuring the likelihood of losing control.

 

[EVNow]

I'm saying they use an inertial measurement unit for vehicle dynamics. I was pointing out that if they just used a simple accelerometer aligned with the direction of travel, the actual deceleration of the car under regen would decrease going downhill, because the gravity component would be added to the regen and the sum would be limited to 0.2g.

Right. If we are going downhill at constant speed because regen & gravity cancel each other
- if the force is measured by speed change, it would be zero
- if the force is measured by an accelerometer - it would be 0.2g (let's assume).

Then, if we brake to make the car stop at the bottom of the hill for a traffic light ...
- if the force is measured by speed change, it would be low deceleration
- if the force is measured by an accelerometer - it would be 0.2g + braking force.

 

[napabill]

Has anyone made the grade and received FSD 10.1? Where will that be reported?

 

[jebinc]

There are some 30% grades in San Francisco (though not, heh, Tampa, where your profile claims to be). Seems like it should be a straightforward thing for some enterprising geek to go out and test.

But I agree, there's no need for accelerometer input here. Just reading instantaneous speedometer output is more than sufficient. Turning rate gets harder though, depending on how much precision they can read the wheel position. There, I might believe that e.g. "turning hard on a banked track" would count for less lateral acceleration. But then... it probably should, if it's measuring the likelihood of losing control.

I found the worse move for aggressive turning dings is being stopped at a corner, then nailing it to merge into traffic when turning left or right.

 

[EVNow]

I hope I’m not repeating the question,

Does anyone know if the weight of a bad drive is by the single drive miles or the entire day? Just wondering if I should cancel my plans
I was at at 99 up until yesterday, but todays score is 49 .

What are the problems you got ? Search through this thread to figure out how to correct / improve them - and do those today.

 

[yurmix]

Drive a lot today! Once tomorrow comes, you're screwed if you don't get today's score up. BOHICA, as they say.... @2021plaid

It just dinged

What are the problems you got ? Search through this thread to figure out how to correct / improve them - and do those today.

I had a forward collision warning pop up. My total average went from 99 to 98.
I fear if I drive more today it will account higher weight.

 

[tivoboy]

I hope I’m not repeating the question,

Does anyone know if the weight of a bad drive is by the single drive miles or the entire day? Just wondering if I should cancel my plans
I was at at 99 up until yesterday, but todays score is 49 .

FCW out of nowhere? I had pretty much the same drop from 100 to 53 during a 4 mile drive and an 11 mile day. It was day 1 essentially, so didn’t know about boinking the system yet. Boinking is your friend.

I spent the next 11 days trying to get it back up, but only made it to 95 yesterday. It would take 700 miles to get it to 99 in three days, so i said spamoni

 

[PDX-Y]

I fear if I drive more today it will account higher weight.

That's not how it works. The Forward Collision Warning metric is expressed as warnings per thousand miles of (non-autopilot, I believe, though I'm not sure) driving. Driving more without safety violations is the only way to increase your score. The daily metrics are for your information only (e.g. "who's fault was that?")