Frustrated with FSD timeline

[whitex]

The EAP software has taken longer than expected and not quite on par with AP1, but Tesla hasn't misrepresented anything. Saying something is expected at a certain time is not a promise or guarantee.

Tesla is playing with words, which yes, can be interpreted literally in such a way that they really are obligated to nothing other that to release some undefined software every so often. It's the letter vs. the spirit of their statements. On the flip side, if you like literal interpretations, my purchase contract doesn't state "United States dollars issued by US Federal Reserve", so do you think they would accept my payment in Taiwan dollars, you know, since the contract never said exactly which dollars?

 

[AnxietyRanger]

The EAP software has taken longer than expected and not quite on par with AP1, but Tesla hasn't misrepresented anything. Saying something is expected at a certain time is not a promise or guarantee.

I am not convinced that would hold water, though.

While Tesla did use the word expected and thus some delay could be reasonable, they did indicate in their original wording (which 2016 made orderers saw when selecting EAP) that a vastly different EAP capability is what was expected in December 2016.

It seems quite possible Tesla knowingly misled and that would change everything. Saying something is expected (to make a sale) when it actually is not expected or possible at all would be misleading...

Even Elon Musk added to this with his claim that already in December 2016 EAP will be better than AP1.

Here we are in April 2017 and EAP has not even reached AP1 parity let alone fit the descriptions Tesla gave in October 2016...

 

[AnxietyRanger]

Lol. I never changed the link to the MRR, it was always the same link. And you proved my point, the AP2 radar has a 160m range like the MRR, so it is likely the MRR which has the elevation data (AKA it is 3D)!

LRR has a range for 250m, so it is unlikely to be that. I hate to break it to you, but the older LRR4 has an elevation antenna too, just like the MRR.
http://q-www-ir7.webspace.bosch.com...roduct_data_sheet_fernbereichsradarsensor.pdf

So all of the data is supporting me. Not matter if they used the MRR or the older LRR4, it has an elevation antenna.

As for the claim that AP1 and AP2 uses the same radar, none of us know that.

Interesting sub-conversation.

I must add that had the radar improved between AP1 and AP2 I would find it likely that

a) Tesla would have told us about it, they did state all other improvements including improvements to ultrasonic and rear camera (though of course to be exact the optics of latter were upgraded earlier)

b) Tesla made the point cloud claim already on AP1 suggesting capabilities of AP1 radar are same across the board

c) Efforts in b) so late in the game suggest they likely were applicable to both AP1 and AP2 hardware

All in all, the reasonable conclusion is that likely both @Bladerskb and @stopcrazypp have both valid and invalid points. I would say stopcrazypp is likely reaching when assuming AP2 features an upgraded radar (while possible nothing points to it, so I am not sure it should be used as an argument yet) and Bladerskb is too dismissive of the point cloud potential of Tesla's radar in general.

As for the point about AP2 braking because of overhead signs, already Elon said AP1 could get false positives from them (hence it did not break for the fatal crash) so IMO that does not point towards a new radar either.

That said, I guess all agree the point cloud of radar is nowhere near the quality of Lidar. I believe even Elon described it as coarse. Elon also said they used temporal smoothing to achieve it, so time definitely has an element.

A fair summary?

 

[stopcrazypp]

I am not convinced that would hold water, though.

While Tesla did use the word expected and thus some delay could be reasonable, they did indicate in their original wording (which 2016 made orderers saw when selecting EAP) that a vastly different EAP capability is what was expected in December 2016.

It seems quite possible Tesla knowingly misled and that would change everything. Saying something is expected (to make a sale) when it actually is not expected or possible at all would be misleading...

Even Elon Musk added to this with his claim that already in December 2016 EAP will be better than AP1.

Here we are in April 2017 and EAP has not even reached AP1 parity let alone fit the descriptions Tesla gave in October 2016...

I'm not a lawyer, but I think what it would fall on if this was suing under breach of contract vs false advertising. The "expected" word is unlikely going to qualify the claim as a hard promise for breach of contract, but it may still be considered misleading under advertising laws. Also, the interpretation of the statement (is it interpreted as that the update includes all the listed features, or just part of it) is going to be play a big part.

 

[stopcrazypp]

Interesting sub-conversation.

I must add that had the radar improved between AP1 and AP2 I would find it likely that

a) Tesla would have told us about it, they did state all other improvements including improvements to ultrasonic and rear camera (though of course to be exact the optics of latter were upgraded earlier)

b) Tesla made the point cloud claim already on AP1 suggesting capabilities of AP1 radar are same across the board

c) Efforts in b) so late in the game suggest they likely were applicable to both AP1 and AP2 hardware

All in all, the reasonable conclusion is that likely both @Bladerskb and @stopcrazypp have both valid and invalid points. I would say stopcrazypp is likely reaching when assuming AP2 features an upgraded radar (while possible nothing points to it, so I am not sure it should be used as an argument yet) and Bladerskb is too dismissive of the point cloud potential of Tesla's radar in general.

As for the point about AP2 braking because of overhead signs, already Elon said AP1 could get false positives from them (hence it did not break for the fatal crash) so IMO that does not point towards a new radar either.

That said, I guess all agree the point cloud of radar is nowhere near the quality of Lidar. I believe even Elon described it as coarse. Elon also said they used temporal smoothing to achieve it, so time definitely has an element.

A fair summary?

About AP1 vs AP2 having the same radar, I agree you can say it is likely that they do from circumstantial evidence. However, you can't say for certain unless you have hard evidence, none of which is present (I don't believe anyone had actually checked the part numbers or actually took the radar sensor out to look). So we shouldn't be presenting that this as fact. We don't even know if it's the same supplier. Tesla had also never mentioned the range of the AP1 radar, so we don't have anything to cross reference (while for the ultrasonic sensors Tesla previously said 16ft for AP1, and now 8m/26ft for AP2).

Also, if the AP1 used a LRR which has a 250m range, and the AP2 switched to the MRR which has a 160m range, Tesla probably would not want to mention that.

Yes, the resolution of the point cloud the current mass produced automotive radar can generate is unlikely to be the resolution of some of the lidar sensors in use right now in prototypes (I use this wording, because it is possible to design a radar sensor superior in resolution to a given lidar sensor; it's not something inherent to radar as is frequently suggested).

However, the point was if Elon was outright lying when he said a point cloud can be made with radar. So far I have not seen evidence that it's impossible to make a point cloud from automotive radar. They can output the distance, azimuth angle, and elevation angle, giving 3 dimensions that can be plotted in a point cloud.

 

[lunitiks]

About AP1 vs AP2 having the same radar, I agree you can say it is likely that they do from circumstantial evidence. However, you can't say for certain unless you have hard evidence, none of which is present (I don't believe anyone had actually checked the part numbers or actually took the radar sensor out to look).

I should've stopped this discussion long ago. I've checked. They are the same part numbers in Teslas parts catalogue.

And here are the Bosch specs:
http://www.bosch-mobility-solutions...data-sheet-mid-range-radar-sensor-(mrr)-2.pdf

 

[stopcrazypp]

I should've stopped this discussion long ago. I've checked. They are the same part numbers in Teslas parts catalogue.

And here are the Bosch specs:
http://www.bosch-mobility-solutions...data-sheet-mid-range-radar-sensor-(mrr)-2.pdf

Ok, thanks, that settles it then, both AP1 and AP2 is using the newer MRR with the elevation antenna.

It would have been helpful if you brought this up much earlier, as there had been constant speculation that Tesla was using an older LRR sensor (perhaps even an older 2D version without an elevation antenna).

 

[AnxietyRanger]

However, the point was if Elon was outright lying when he said a point cloud can be made with radar. So far I have not seen evidence that it's impossible to make a point cloud from automotive radar. They can output the distance, azimuth angle, and elevation angle, giving 3 dimensions that can be plotted in a point cloud.

Yes, I don't think he was lying on that one. Agreed.

I was just trying to summarize what angles from your and Bladerskb's messages rang true to me. I am happy to agree with your "I agree you can say it is likely that they do from circumstantial evidence" on the radars (based on what we knew before @lunitiks message).

I think the real reality here is that all of us, we are learning (hopefully!) something every message we read and write. The communal knowledge is constantly evolving. Even Bladerskb who may well be an expert in neural networks, but not always do we know everything about everything. He made a wrong assumption about the radar used in AP1, on that we agree, though he was right it was the same radar in both. None of us really, looking back at the recent exchanges, knew exactly - it seems - what the radar was capable until @lunitiks came in and settled the score. Assuming he is right. Live and learn.

I thus agree even a single "frame" (yeah, I know thinking in those terms is misleading but bear with me) of the AP1 and AP2 radar can produce a 3D image - this was likely and now the specs would seem to confirm it too - and the time element is then used as a fourth dimension (sort of) to make sense of that frame, because of radar's tendency to misrepresent certain things (e.g. metal, certain shapes) in a given single moment of time. So that single "frame" is "smoothed" through comparing several moments in time. Might that be an accurate description?

Bosch even has an image showing how overhead signs would collide with the radar beam on the MRR front radar used by Tesla AP1 and AP2:

Thank you @lunitiks for the additional data point.

 

[AnxietyRanger]

I'm not a lawyer, but I think what it would fall on if this was suing under breach of contract vs false advertising. The "expected" word is unlikely going to qualify the claim as a hard promise for breach of contract, but it may still be considered misleading under advertising laws. Also, the interpretation of the statement (is it interpreted as that the update includes all the listed features, or just part of it) is going to be play a big part.

I am not so much even concerned with the legalese - like you, I think the nuanced interpretations of which probably are left to the lawyers. Maybe what you say applied, maybe something else.

What I am concerned mostly with is what I think a reasonable layperson would have thought of Tesla's sales text for the Enhanced Autopilot in October 2016. I am concerned about what a reasonable bystander would find right.

And I think the reasonable layperson, paying reasonable attention to Tesla's sales pitch and knowning what they should reasonably know when assessing that type of information, would have come - in October/November/December 2016 (pre release) - to the conclusion that EAP would be far superior to AP1 in the form expected for December 2016. They would have reasonably expected four active cameras, auto lane changes and all that - or at the very least, were they listening to Elon instead of the Design Studio text, a product superior to AP1 in December 2016.

The AP1 parity talk, that came only after it was found out EAP was nowhere near ready as advertised. The original goal for December 2016, as we were sold it, was EAP - superior to AP1 - expected in December 2016. In an old switcharoo it later morphed into an AP1 parity quest.

Nothing in my mind would have prepared that reasonable layperson to a prolonged, quarter or two's chase after AP1 feature parity instead of getting EAP. That is not how it was sold at that time. Some reasonable delay from December 2016 is one thing, that might have been understandable given the wordings (say, EAP being late by a month or two), but what the wordings did not say was that what was being expected for December 2016 wasn't EAP at all, but a subset of AP1 that originally used only one camera.

Venturing into the legalese is a slippery slope. Lawyers can worry about that. I think we should mostly be concerned here what is reasonable and right. And I have come to the conclusion that Tesla failed in that test.

 

[JonG]

I am not so much even concerned with the legalese - like you, I think the nuanced interpretations of which probably are left to the lawyers. Maybe what you say applied, maybe something else.

What I am concerned mostly with is what I think a reasonable layperson would have thought of Tesla's sales text for the Enhanced Autopilot in October 2016. I am concerned about what a reasonable bystander would find right.

And I think the reasonable layperson, paying reasonable attention to Tesla's sales pitch and knowning what they should reasonably know when assessing that type of information, would have come - in October/November/December 2016 (pre release) - to the conclusion that EAP would be far superior to AP1 in the form expected for December 2016. They would have reasonably expected four active cameras, auto lane changes and all that - or at the very least, were they listening to Elon instead of the Design Studio text, a product superior to AP1 in December 2016.

The AP1 parity talk, that came only after it was found out EAP was nowhere near ready as advertised. The original goal for December 2016, as we were sold it, was EAP - superior to AP1 - expected in December 2016. In an old switcharoo it later morphed into an AP1 parity quest.

Nothing in my mind would have prepared that reasonable layperson to a prolonged, quarter or two's chase after AP1 feature parity instead of getting EAP. That is not how it was sold at that time. Some reasonable delay from December 2016 is one thing, that might have been understandable given the wordings (say, EAP being late by a month or two), but what the wordings did not say was that what was being expected for December 2016 wasn't EAP at all, but a subset of AP1 that originally used only one camera.

Venturing into the legalese is a slippery slope. Lawyers can worry about that. I think we should mostly be concerned here what is reasonable and right. And I have come to the conclusion that Tesla failed in that test.

I think you can go further, the wording states EAP adds new capabilities to the existing AutoPilot driving experience. It's entirely plausible that stepping out of an AP1 car and reading that you'd think that EAP already had all the AP1 capability and it's only the new features that are subject to legislation etc.

 

[Kant.Ing]

The pricing could be the last concern. At that time (even now), it was like the EAP $5,000 is helping Tesla making the transition and buying only the hardware without software, as AP1 and AP2 on text and on twitter can/will do almost the same tasks, but $5,000 hardly justifies the cost delta between Mobileye and nVidia hardware, and 3 additional cameras.

The AP1's $2,500 not only gave us a (working) hardware but also a much much better working software.

Wonder if people will feel better on a $2,499 EAP.

 

[JeffK]

Sorry, no reasonable person will see it this way. Only an apologist. The fact that AP2 isn't even equal to AP1 is proof, considering I paid for the "enhanced" version of the AP1 car I test drove - not to be a beta test dummy.

You think I would have paid a premium for an AP2 car if they told me it would actually be WORSE than a used AP1 car? I could have saved $60,000 and had a more capable car.

You'd think you would have read the disclaimer first... Everyone who did even the slightest bit of research knew that AP2 would not have the same features as AP1 right away. Here's a direct quote from Elon:

...the feature-set initially will be disabled say well at least for the first few months. The cars with hardware 2.0 which is full autonomous suite will actually have fewer features than cars with Hardware 1.0.

He later says that expects feature parity in Dec 2016 which we know he was wrong about, but it doesn't mean he promised it.

He also says:

I feel pretty good about this goal is that we will be able to demonstrate a demonstration drive of our full autonomy all the way from LA to New York. So basically from home in LA to let’s say dropping you off in Times Square, NY and then having the car parking itself by the end of next year (2017) without the need for a single touch including the charger.

"I feel pretty good about..." <-- also not a promise. It might be the end of 2017 it might be the end of 2018, who knows. The software will be done when it's done. You can't make a guaranteed promise of when the complete full self driving suite is going to be delivered.

I feel bad that you're sore about not doing research ahead of time or having unrealistic expectations.

 

[JeffK]

The pricing could be the last concern. At that time (even now), it was like the EAP $5,000 is helping Tesla making the transition and buying only the hardware without software, as AP1 and AP2 on text and on twitter can/will do almost the same tasks, but $5,000 hardly justifies the cost delta between Mobileye and nVidia hardware, and 3 additional cameras.

The AP1's $2,500 not only gave us a (working) hardware but also a much much better working software.

Wonder if people will feel better on a $2,499 EAP.

Granted the Nvidia hardware costs significantly more than the MobilEye chip, the consumer price has more to do with Model 3 ramp up and future value to the customer when it's full potential is realized vs actual cost of the equipment.

 

[AnxietyRanger]

You'd think you would have read the disclaimer first... Everyone who did even the slightest bit of research knew that AP2 would not have the same features as AP1 right away. Here's a direct quote from Elon:

He later says that expects feature parity in Dec 2016 which we know he was wrong about, but it doesn't mean he promised it.

I feel bad that you're sore about not doing research ahead of time or having unrealistic expectations.

That quote - reasonably - simply referes to the fact that initially AP2 cars shipped without EAP. It was not news that initially cars would ship without AP or EAP. But what Tesla.com Design Studio said was an Enhanced Autopilot was expected to come in a December 2016 update (i.e. reasonable interpretation: cars delivered in October, November and early December would be without and then it would come, if not in December 2016 than reasonably soon afterwards). Even Elon himself said EAP would be beyond AP1 starting in December 2016.

A small delay would have been one thing, but instead they shipped a completely different software feature-set instead. None of that early EAP talk, in hindsight, was representative of what occurred. Instead we are still seeking AP1 parity and EAP is nowhere in sight. Obviously EAP was not going to be beyond AP1, let alone the four camera auto-lange changing thing Design Studio talked about, in December 2016. But Tesla did not tell us this.

He also says:

"I feel pretty good about..." <-- also not a promise. It might be the end of 2017 it might be the end of 2018, who knows. The software will be done when it's done. You can't make a guaranteed promise of when the complete full self driving suite is going to be delivered.

Irrelevant as Elon is talking FSD there. We are not talking about FSD but EAP now in these latest messages, including @oktane in the message you replied to. FSD, most agree, was only said to be coming at some future date later. But of EAP they said this - this is what the customer ordering EAP read:

 

[JeffK]

That quote - reasonably - simply referes to the fact that initially AP2 cars shipped without EAP. It was not news that initially cars would ship without AP or EAP. But what Tesla.com Design Studio said was an Enhanced Autopilot was expected to come in a December 2016 update (i.e. reasonable interpretation: cars delivered in October, November and early December would be without and then it would come, if not in December 2016 than reasonably soon afterwares). Even Elon himself said EAP would be beyond AP1 starting in December 2016.

None of that, in hindsight, was representative of what occurred. Instead we are still seeking AP1 parity and EAP is nowhere in sight. Obviously EAP was not going to be beyond AP1, let alone the four camera auto-lange changing thing Design Studio talked about, in December 2016. But Tesla did not tell us this.



Irrelevant. We are not talking about FSD but EAP now in these latest messages, including @oktane in the message you replied to. FSD, most agree, was only said to be coming at some future date later. But of EAP they said this - this is what the customer ordering EAP read:

Again the wording "expected" is not a guarantee, it's an outright guess given available evidence. Unfortunately the English language is subject to interpretation.

Personally, I feel that those people too lazy to wait for EAP on AP 2 might as well sell their cars for an older AP 1 car if it makes them more happy, just don't come crying when EAP is finally more accurate and has more features than AP 1.

Next time you make an expensive purchase, research, research, research. I would feel grateful to be able to afford a high end Model S while I'm not of retirement age.

Instead of disagreeing Sandy Chamois why not come post why you disagree? You too croman

 

[AnxietyRanger]

Again the wording "expected" is not a guarantee, it's an outright guess given available evidence. Unfortunately the English language is subject to interpretation.

Personally, I feel that those people too lazy to wait for EAP on AP 2 might as well sell their cars for an older AP 1 car if it makes them more happy, just don't come crying when EAP is finally more accurate and has more features than AP 1.

My point is: I do not fully believe EAP in the form advertised was really "expected" for December 2016. And if my fear is correct, that would mean Tesla intentionally misled or acted in a manner where a reasonable layperson customer, using reasonable diligence and effort, could not be expected to understand Tesla's messaging realistically.

Look at TMC in October-November 2016. If a lot of enthusiasts understood Tesla wrong on the EAP ramp-up, how could a layperson be expected to get it?

After all, that Design Studio text/image which I keep quoting, is actually quite direct in saying what EAP was expected for December 2016. I keep quoting it, and people keep ignoring it, because it is in actual fact rather damning. If that is what you saw when selecting your options for the car, asking people to listen to the CEO would not be reasonable. Yet not even Elon Musk said EAP would be less than AP1 in December, he actually said it would be beyond AP1 in December 2016... so not even listening to Elon Musk would have helped.

Scepticism and not believing Tesla's words would have helped. But then what does that tell us of Tesla's message in the first place.

Next time you make an expensive purchase, research, research, research.

I have no problem with my personal purchase (well nothing research or AP2 related anyway). I do not consider myself a regular customer anyway. Safe to say, I have analyzed my Model X purchase since 2014 beyond anything that is reasonable. This is about analyzing Tesla's actions and reasonable interpretations of them.

Besides, just because Tesla has a history of missing schedules and whatnot, doesn't mean they shouldn't be asked to do better in the future. After all, it is easy: they could user under-promise.

 

[JeffK]

My point is: I do not fully believe EAP in the form advertised was really "expected" for December 2016. And if my fear is correct, that would mean Tesla intentionally misled or acted in a manner where a reasonable layperson customer, using reasonable diligence and effort, could not be expected to understand Tesla's messaging realistically.

That's what the court case is about. We'll have to stay tuned to see how that comes out. They did get slapped on the wrist regarding the horsepower numbers in Europe a while back and rightly so.

 

[boaterva]

That's what the court case is about. We'll have to stay tuned to see how that comes out. They did get slapped on the wrist regarding the horsepower numbers in Europe a while back and rightly so.

Um, isn't the court case about 'full FSD' which is/was never expected/promised anything for 'the past'? EAP, we can talk about all day long (and have...), but people expecting 'real FSD (level 4/5)', have no proverbial (legal or otherwise) leg to stand on.

 

[JeffK]

Um, isn't the court case about 'full FSD' which is/was never expected/promised anything for 'the past'? EAP, we can talk about all day long (and have...), but people expecting 'real FSD (level 4/5)', have no proverbial (legal or otherwise) leg to stand on.

Both FSD and current capabilities people are calling "unsafe"
Tesla Sued Over ‘Dangerously Defective’ Autopilot Software

There seems to be "confusion" about current capabilities.

 

[BestRadar]

Both FSD and current capabilities people are calling "unsafe"
Tesla Sued Over ‘Dangerously Defective’ Autopilot Software

There seems to be "confusion" about current capabilities.

So if you are in AP and you come up on a slow moving truck and AP does not see it and does not slow down is that not unsafe? Or just confusion on its capabilities?