OFFICIAL BUTTON WATCH

[Knightshade]

Does Tesla compare highway driving of other, modern cars or just overall accident rates though?

Based on Musk's quote, sounds like AP is ready for no driver supervision on the highway at all.

They compare accident rate "Tesla on AP" to "Tesla not on AP but active safety features", "Tesla not on AP active safety turned off", and "All cars period"

The AP miles will tend to skew to being highway miles though.

Tesla Vehicle Safety Report | Tesla

At Tesla, we believe that technology can help improve safety. That’s why Tesla vehicles are engineered to be the safest cars in the world. We believe the unique combination of passive safety, active safety, and automated driver assistance is crucial for keeping not just Tesla drivers and...

www.tesla.com

APs numbers are only that good because it's supervised.

I have generally easy-for-AP roads, and a pretty straightforward almost entirely highway drive to work- and I'd absolutely have had accidents if I'd been reading a book and not ready to immediately take over. Several at a minimum.


it's an awesome system, and makes highway driving 10,000% better- but it's NOT something I'd trust unsupervised.


(and I write that as someone who bought FSD in the first place, back in 2018, with "be able to read a book during highway driving" as my minimum victory condition for the purchase)

 

[qdeathstar]

It isn’t. If all collective Tesla drivers went to sleep on AP on highway most would wake up dead.

yeah, but they don’t close their eyes. He isn’t saying it needs to be level 5.

 

[boonedocks]

Did Elon say hopefully or “hopeless” in his tweet? This is exactly why I did NOT order FSD on my ‘21 Refresh S.

I have believed Elon for 3 full payments for FSD on 3 Tesla’s when AutoPilot is all I have regularly used since 2015 S AP1 & 2017 S & 2017 X & 2019 S

I don’t think anything truly meaningful ESPECIALLY hands free will be available until I am 3 more new Tesla’s in.

 

[AlanSubie4Life]

highway AP is already above 99.999999% reliability

As has been pointed out, with supervision.

Tesla publishes accident rate data on AP every quarter. It's VASTLY lower than without AP (in a tesla) and vastly lower than that again compared to non-tesla cars.

The absolute number for AP on, I won't quibble with (and that's what matters here), even if it is unverifiable. But as has been discussed ad nauseam, the comparative accident rates are impossible to verify, due to massive selection biases, everywhere. We literally have no idea how much safer using AP is, even compared to a Tesla not using AP, since we have no idea what is being compared, exactly.

As has been pointed out, driving with headlights off is safer than driving with them on. While true, no one would make this claim.

NOTE: I think using AP is safer than not using it. We just don't know (I think it probably is, but we have no actual data to support that, and it would be great if we did - Tesla has the data, but unfortunately they have not shared it).

sounds like AP is ready for no driver supervision on the highway at all.

No, he's a salesman, so he just didn't mention the detail that that safety level requires driver supervision. Just as City Streets will.

 

[powertoold]

I am curious though. If highway AP is already above 99.999999% reliability now, why doesn't Tesla remove driver supervision for highway driving? Why don't they announce L3 highway if they have achieved their safety benchmark?

with just basic AP on highway, how many miles do you think it can go in one lane (hypothetically no interchanges) before it gets into an at-fault accident? Without supervision

i think at least 40,000 miles, although there will be moments of uncomfortable braking

 

[run-the-joules]

It isn’t. If all collective Tesla drivers went to sleep on AP on highway most would wake up dead.

Elon would dismiss that by saying we can't prove we're alive right now. It's all a simulation!!!1one.

 

[diplomat33]

with just basic AP on highway, how many miles do you think it can go in one lane (hypothetically no interchanges) before it gets into an at-fault accident? Without supervision

i think at least 40,000 miles, although there will be moments of uncomfortable braking

Honestly, I have no idea. If we are talking easy cruising in just one lane (no lane changes, no interchanges), probably a lot of miles. If we add more complicated scenarios, lane changes, construction zones, etc... it would be less.

 

[diplomat33]

They compare accident rate "Tesla on AP" to "Tesla not on AP but active safety features", "Tesla not on AP active safety turned off", and "All cars period"

The AP miles will tend to skew to being highway miles though.

Tesla Vehicle Safety Report | Tesla

At Tesla, we believe that technology can help improve safety. That’s why Tesla vehicles are engineered to be the safest cars in the world. We believe the unique combination of passive safety, active safety, and automated driver assistance is crucial for keeping not just Tesla drivers and...

www.tesla.com

APs numbers are only that good because it's supervised.

I have generally easy-for-AP roads, and a pretty straightforward almost entirely highway drive to work- and I'd absolutely have had accidents if I'd been reading a book and not ready to immediately take over. Several at a minimum.


it's an awesome system, and makes highway driving 10,000% better- but it's NOT something I'd trust unsupervised.


(and I write that as someone who bought FSD in the first place, back in 2018, with "be able to read a book during highway driving" as my minimum victory condition for the purchase)

Good point. Basically the 99.999999% is meaningless since it is with supervision.

 

[Bladerskb]

with just basic AP on highway, how many miles do you think it can go in one lane (hypothetically no interchanges) before it gets into an at-fault accident? Without supervision

i think at least 40,000 miles, although there will be moments of uncomfortable braking

Not even 100 miles on average if It encountered a tire on the road or any road debris, or park a truck in the lane or car off to the side of the road 15% in the lane.

 

[Cheburashka]

Not even 100 miles on average if It encountered a tire on the road or any road debris, or park a truck in the lane or car off to the side of the road 15% in the lane.

Exactly.

For example, my wife doesn't really care about tech or AP, so I view her an an average Tesla owner that doesn't really care for trying to push it to the limits. She only uses it in a situation where the road is straight and there isn't too much traffic, or traffic is completely backed up. I think this is how most AP users are. AP handles so many situations poorly that people just know to only use it in certain scenarios, which is why it's technically "safe" to begin with.

 

[powertoold]

Not even 100 miles on average if It encountered a tire on the road or any road debris, or park a truck in the lane or car off to the side of the road 15% in the lane.

You’re right, but I don’t think it’s less than 100 miles on avg. Then again, we’re not debating any facts.

I just want to point out Elon’s logic here. I think he’s saying that for what Tesla AP is designed for on highway, it has a very low likelihood of accident. But, Tesla AP on highway isn’t designed to avoid road debris or construction.

City FSD is designed for these aspects.

So when I say I think basic AP on highway wouldn’t be in an at fault accident for at least 40,000 miles without supervision, I’m talking about within its design specifications.

 

[Cheburashka]

You’re right, but I don’t think it’s less than 100 miles on avg. Then again, we’re not debating any facts.

I just want to point out Elon’s logic here. I think he’s saying that for what Tesla AP is designed for on highway, it has a very low likelihood of accident. But, Tesla AP on highway isn’t designed to avoid road debris or construction.

City FSD is designed for these aspects.

So when I say I think basic AP on highway wouldn’t be in an at fault accident for at least 40,000 miles without supervision, I’m talking about within its design specifications.

Doubt it. I have a highway near my house with steep-ish turns and AP hogs the side of the lane so often that I'd get a collision at least once every 100 miles of driving it.

Look, they aren't even using AP for the Vegas loop tunnel thing and the max speed there is a measly 30mph with nothing but Teslas inside.

 

[Cheburashka]

Also I've seen AP totally ignore empty car carrier trailers like this:

 

[rxlawdude]

I guess the only positive is I won't need to be constantly checking here for the FSD button. At least until, hopefully, June (year indeterminate).

 

[Cheburashka]

hahaha! We have a new timeline here folks, and it's even worse than today's earlier timeline of "hopefully", we're now ASPIRATIONAL

 

[boonedocks]

Meanwhile Ford is updating Bluecruise enabled cars to be hands free on 100k+ miles of roadways INCLUDING 90 miles of my daily commute. Come on Tesla at least give us more than 13.987 seconds of hands free on expressways!!!

https://twitter.com/x/status/1382283651073724417

 

[Knightshade]

Fords system is hands free because that have a dedicated set of driver monitoring hardware insuring your eyes are on the road.

Teslas is not because their primary driver monitoring system is a torque sensor on the steering wheel.


Tesla IS trying to back-port the never-meant-for-the-job robotaxi selfie cam as a driver monitor currently... it's not nearly as capable as Ford of Caddys system, but much like "vision no radar" they might just decide it's "good enough" and then with one software push to turn off the torque sensor your Tesla is hands-free too-- and on a lot more roads (and with far greater capabilities) compared to Ford.

 

[Dan D.]

I am curious though. If highway AP is already above 99.999999% reliability now, why doesn't Tesla remove driver supervision for highway driving? Why don't they announce L3 highway if they have achieved their safety benchmark?

But what Elon said is "Production Autopilot [probability of no injury] is already above [99.999999%] for highway driving"

I think what he is saying is that there have been Y injuries over the last X million miles - for a very small number Y. But that's a statistic not a probability. I don't see how he gets a probability out of a statistic.

If I have thrown heads 9 times of the last 10 throws that's a result of 90% heads. 90% is not the probability of heads, it's still 50%.

 

[boonedocks]

Fords system is hands free because that have a dedicated set of driver monitoring hardware insuring your eyes are on the road.

Teslas is not because their primary driver monitoring system is a torque sensor on the steering wheel.


Tesla IS trying to back-port the never-meant-for-the-job robotaxi selfie cam as a driver monitor currently... it's not nearly as capable as Ford of Caddys system, but much like "vision no radar" they might just decide it's "good enough" and then with one software push to turn off the torque sensor your Tesla is hands-free too-- and on a lot more roads (and with far greater capabilities) compared to Ford.

Mmhhmmmm

 

[aronth5]

Knowing Elon's communication skills when it comes to timelines are difficult to fathom all I care about now is the release of the limited beta for V9 so new videos are available. The general release of the button is less important. I'll make my judgement on the value of FSD once the beta testers provide their feedback. I'll ignore all the whining in this thread until then.