Bummer for those who paid for self drive.

[ebmcs03]

Just like the v9 dashcam footage where the red Model 3 t-bones the white Honda.

Is there a link? I want to see how the dash cam performs

Found it:

 

[Hitman007]

Spare us the (fake) deep concern and condolences @Hitman007 .

I'll make this really simple for you and everyone else:

FSD worth = $30,000
Incremental FSD worth over the next few years = $3,000
Locked in hardware upgrades = Priceless

no need to regurgitate a Mastercard commercial....

 

[brianman]

Does what you paid for FSD transfer over to the new car?

No.

 

[Dana1]

question for those who got FSD. I realized that you will be HW3 upgrade for free, but what if the ability for FSD doesn't arrive before you want to sell or trade in the car for a new model? Does what you paid for FSD transfer over to the new car?

If the potential buyer values it, yes. Even on ICE cars you usually find the cars with more features and larger engines have higher resale values.

 

[Dana1]

I agree with your conclusion (FSD will take a long time to develop), but not the reasons that you state.

You say: "It is just so limited and can't see fast motorcycles lane splitting or 3 lanes down that a car is going to slam into you if you keep the same vector."
It absolutely can *see* the motorcycle and the car 3 lanes down, it just doesn't "understand" (I use the term loosely, because AI is not really "intelligent" per se) those situations. This is a matter of training and compute capacity.

"AP can't scan and predict." - absolutely untrue. The AP can predict. The problem with using AI to predict is that it can also generate false positives (i.e., it's easy for it to think a given car is going a certain place when it isn't). Tesla's current implementation is (properly, IMO) very conservative - by primarily being reactive it avoids a lot of situations where it may mistakenly take action on something it thinks might happen, but doesn't.

A bit of the problem of prediction is evident in the false braking events people report when approaching overpasses - it's easy for the AI to misinterpret inputs (this is known as "overfitting" in AI training terms).

"This is why most FSD cars like Google and Cruise gets rear ended." - FSD cars get rear-ended because this is how the most classic human error exhibits itself - inattentiveness. They get rear-ended because they get hit by people texting, fiddling with the radio, doing their makeup, etc. This is also why non-FSD cars constantly get rear-ended. I've seen no evidence to suggest that FSD cars get into accidents different from the accidents human drivers get into. Note that by most reports FSD cars *cause* very few accidents - the human element is still the weak link here.

IMO, the reality here is that true FSD requires not only a "good" solution - it requires a "great" solution. FSD will be given little benefit of the doubt, so it will be held to a much higher standard than humans will be held to. This means that there is a ton of situations the FSD will handle cleanly before gaining adoption.

It's one thing to come up with a technology that solves 99% of the problems - it's much, much harder to come up with the technology that solves 100% of them. I think there are several vendors at the 99% stage - but not many approaching 100%.

Agree. The plastic bag problem. I think the technology will develop quickly, as it did for ships, but may require a driver observer just like ships. The human does nothing for weeks on end until that 1% problem occurs and they have to step in.

 

[MXWing]

If the potential buyer values it, yes. Even on ICE cars you usually find the cars with more features and larger engines have higher resale values.

Yes YMMV for sure. However prepaid FSD doesnt exist on any other ICE car or unlimited supercharging.

Those are call options that vary in value until the vehicle is gone.

 

[Foxtrotter]

No bummer here! I bought FSD on our two model 3's as a kickstarter/challenge/support for/to Tesla and Musk. I'm 76 and this may well be my last car. If the new computer helps EAP work better with its faster processing that would be enough to justify it for me. Any help I can get, even if not "full self driving" will be deeply appreciated. We drive a lot. 14,000 miles on the two cars that we have owned for 5 and 3 months. Not at all impaired in driving, just not as good as I was when younger. EAP is just a godsend and if more computing power will improve it that all I ask for.

 

[Reeler]

Here is my guess on why they pulled it. I think Tesla finally realized that they will need far more hardware to make FSD work. They will have to substantially rework the sensor suite on the cars sold with this feature. To limit their rework, they stopped offering the feature.

I don't think it will just be a board swap. I think they will need to add more sensors and/or redundancy. AP3.0 will include these and Tesla will not offer the feature again until the cars are upgraded. They will offer a retrofit, but I expect it will be $5K or more and may take a while to actually deliver those upgrades.

 

[ohmman]

mod note: four posts moved to snippiness.

 

[adaptabl]

So what does that mean? That FSD causes only 10,000 or so fatalities per year on US roads rather than the far greater number caused by human drivers? FSD can be pretty awful and still be better than humans. That's basically Elon's reasoning when he says that it's immoral not to deploy it as soon as it's better than humans.

Being killed by a computer is something the public will not accept. I agree it may be safer but I do not see the public accepting a machine killing people. It would have to be 100% safe and that is unlikely to ever happen.

 

[Hitman007]

FSD is not coming. Government has no plans to allow it and only big companies can test on public roads. Which means, only big companies can operate it if allowed. Next step is Govn't opening up testing for personal vehicles which could be 10+ years and probably limited permits that are 1st come 1st serve. Also, getting all 50 states aligned will take a long time. Maybe it can go FSD from CA to AZ, but once you get into TX, you have to drive manually. I don't see how cameras alone will get Govn't approval for FSD either. It is just so limited and can't see fast motorcycles lane splitting or 3 lanes down that a car is going to slam into you if you keep the same vector. Just like the v9 dashcam footage where the red Model 3 t-bones the white Honda. This happens all the time in LA and as a driver I can predict and slow down at each gap knowing there is a high percentage someone will shoot the gap. AP can't scan and predict. It will just go the speed limit and slams on the brake. This is why most FSD cars like Google and Cruise gets rear ended.

I think you are correct.

 

[Dana1]

Being killed by a computer is something the public will not accept. I agree it may be safer but I do not see the public accepting a machine killing people. It would have to be 100% safe and that is unlikely to ever happen.

Y’all realize computers are already controlling drillships, pipelaers, OSVs, Dive vessels, yaddah, yaddah, and have been for forty years. Dynamic Positioning. There are still accidents, some fatalities, tech keeps improving and governments seem fine with it.

 

[lolder]

I think majority of customers who ordered this feature understands that the feature maybe a few years out or at least understands the stipulation.

Funny I was debating this feature when I ordered my car and the sale guy who came to my house with the car for a test drive told me to hold off. He said don’t spend the money until it’s actually release and the government approves it. So I decided to not pay upfront and not pay the taxes for it.

I’d definitely will get it once it is released out of beta etc but by then I may have another Tesla model y. Lol.

Why would you have another Tesla in a few years? The whole point of these cars is that they do not get obsolete.

 

[lolder]

Thanks! Glad there is a start but when will all 50 states agree to this or even just California.



No doubt the tech will be better but my Model 3 is using 2018 tech. Just a CPU swap will not make it FSD. I believe more hardware are needed to satisfy the government. Look what Cruise and Google are using. They have a ton of hardware and still not 100% perfect.

The M3 hardware is already designed to be redundant and "fail safe" and maybe even "fail-operational" like aircraft automatic landing systems. The braking is redundant ( and has been for decades ) and the electric power steering appears to have two actuators. The Tesla cameras have the potential for much more data input than the rotating lidars of the competition. At 64 per car just imagine how many lasers are flashing around a 50 car intersection. Tesla's plan is to teach the car to "see". A CPU update in the AP computer and software that sees are the only things left. Musk acknowledges the risk in his approach. I wouldn't second guess him.

 

[sperkin]

Is there a link? I want to see how the dash cam performs

Found it:

Thanks for finding the video. I'm on my phone so it was harder to find and link. In this situation, a driver has a wider peripheral vision and can see something creeping up. This driver relied too much on AP or was texting. This dash cam is a long lens so objects are closer than they appear. A human eye would have seen this and had time to react. This situation happens all the time in LA, people trying to squeeze in on a left turn and force their way in.

There are many situations that FSD can't account for because people don't follow rules. EAP or FSD will need to predict what every object can perform at any given time to be safe. As a driver you can tell when someone look left to check their blind spots that have a high chance that they will probably cut you off soon. Most drivers drive like A$$ and speeds up to close the gap. AP needs to speed down and give some room so other driver has a safe lane change. AP cameras don't see what other drivers are doing. Many time I see a driver texting, I know not to change lanes in front of them. I choose to go behind them instead.

FSD also need to assess risk. Hit the baby stroller in the middle of the road (cause mom was jaywalking) or turn the car to hit a parked car knowing there is no time to stop to avoid the person.

I just don't have a lot of confidence since I can't see test footage or understand where Tesla is at. Just from my perspective and personal experiences I don't think FSD is as simple as a CPU swap. FSD will probably be like a 13 year old kid trying to drive with experience from playing Froza.

After all this said, I still believe Tesla has the best AP on any production car on the market and it's the only reason why I bought a Tesla. Both my wife and I used EAP 100% of the time on the freeway. She was against Teslas at first but after 1 commute to work with EAP she fell in love with Tesla.

Hope it gets better and FSD comes soon. I will upgrade both cars at $5k each if I can sleep on the way to work. Also, waiting for a day where I can walk to grocery shopping and call my car to come pick me up so don't need to walk in home carrying all the food (I like walking down the mountain but not back up).

 

[Knightshade]

I just don't have a lot of confidence since I can't see test footage or understand where Tesla is at. .

Here's some. From what Tesla already had working in 2016 with the initial AP2 hardware.

Full Self-Driving Hardware on All Teslas

 

[sperkin]

The M3 hardware is already designed to be redundant and "fail safe" and maybe even "fail-operational" like aircraft automatic landing systems. The braking is redundant ( and has been for decades ) and the electric power steering appears to have two actuators. The Tesla cameras have the potential for much more data input than the rotating lidars of the competition. At 64 per car just imagine how many lasers are flashing around a 50 car intersection. Tesla's plan is to teach the car to "see". A CPU update in the AP computer and software that sees are the only things left. Musk acknowledges the risk in his approach. I wouldn't second guess him.

Lots of smarter people at Tesla than me. Hope they can give the car better vision then a human and get FSD out soon, but just for example, everyone says the woman hit by Uber was walking across the street in an area not lit by light. Robot cars shouldn't need lights to see. Right now I think all Model 3s needs light to see. Can they go FSD in pitch dark? There are areas where I grew up had no street lights and lots of deers.

Here is another example where I think they need to solve before FSD can be approved.

What if a bunch of kids run across a street that has a speed limit of 50mph? They see a Tesla coming and stopped in the on coming side of the road, but their unleashed puppy didn't. Will FSD slow down to reduce impact when hitting the puppy or slam on the brake and let the car behind slam into the Tesla? Will the TeslaThe drive behind has a chance of not wearing his seat belt and could get hurt. What will FSD do?

 

[PessimiStick]

What if a bunch of kids run across a street that has a speed limit of 50mph? They see a Tesla coming and stopped in the on coming side of the road, but their unleashed puppy didn't. Will FSD slow down to reduce impact when hitting the puppy or slam on the brake and let the car behind slam into the Tesla? Will the TeslaThe drive behind has a chance of not wearing his seat belt and could get hurt. What will FSD do?

Try to stop for the puppy, accelerate if about to be rammed, would be my guess. That's what I would train it to do at any rate. But it's not like you program these scenarios directly anyway. The car will do whatever it thinks is the safest option.

 

[crackers8199]

This is great for those that locked in FSD at $3k.

not really, those of us who didn't buy it should still have the option to buy at $4k or $5k (depending when they ordered). it's still available in your account as an autopilot upgrade.

 

[crackers8199]

Here is my guess on why they pulled it. I think Tesla finally realized that they will need far more hardware to make FSD work. They will have to substantially rework the sensor suite on the cars sold with this feature. To limit their rework, they stopped offering the feature.

I don't think it will just be a board swap. I think they will need to add more sensors and/or redundancy. AP3.0 will include these and Tesla will not offer the feature again until the cars are upgraded. They will offer a retrofit, but I expect it will be $5K or more and may take a while to actually deliver those upgrades.

i said this in another thread, but it bears repeating: if they try to do this, they will have the mother of all class action lawsuits on their hands. they advertised the car as having all the hardware needed to be fully autonomous, with the ability to add the software later for $4k or $5k (depending on when you ordered) to make that happen. to go back now and say "sorry, it costs $15k (or whatever price it may be) to add the necessary hardware to take care of this in addition to the software" would mean they sold the car initially based on blatant false advertising and under false pretenses.

it's not the customer's fault that elon over-sold the vehicle...it's elon's. legally i can't imagine how they could get out from under allowing any of us who didn't buy FSD at purchase to upgrade whatever hardware is necessary just by purchasing FSD for the $5k we agreed to when we ordered the car.