FIRMWARE UPDATE! AP2 Local road driving...and holy crap

[Bet TSLA]

When Chris Lattner, the new software engineer from Apple was hired a month ago, he described autonomous driving: "It is also a very, very hard technology problem." Notice the use of two 'very's' by someone who really knows what he is talking about.

I don't know of any reason to believe that Chris Lattner knows anything about autonomous driving software or AI. He was a compiler and developer tools guy at Apple. I'm sure he'll learn fast, and I'm also sure that Tesla has many people who are experts in the important areas, but he's not "someone who really knows". Or wasn't when he said that anyway.

Elon, on the other hand, has said that autonomous driving is pretty much a "solved problem".

 

[stefan.bc]

You should always know whether someone is behind you.

Yes, you should, that doesn't mean you always will or can. The autopilot's job is to relieve some of the stress after a few hours of driving not increase it, and help you in situations where a lot of things are happening and you might not be able to react in time. It should also help beginners drive more safely, not the other way around.

And considering how bad the visibility is in my MX(can't say for S) I am not sure I get your point at all.

 

[montefuego]

I don't know of any reason to believe that Chris Lattner knows anything about autonomous driving software or AI. He was a compiler and developer tools guy at Apple. I'm sure he'll learn fast, and I'm also sure that Tesla has many people who are experts in the important areas, but he's not "someone who really knows". Or wasn't when he said that anyway.

Elon, on the other hand, has said that autonomous driving is pretty much a "solved problem".

So, you don't think that Chris investigated the new job before he left his old, and probably higher paid one? I think you are incredibly naive about a guy like Chris and how he operates. He is not 'jumping into the deep end' without first knowing a lot about what is in that water. He is not being hired as an intern, but as the guy who will led the team.

 

[Bet TSLA]

So, you don't think that Chris investigated the new job before he left his old, and probably higher paid one? I think you are incredibly naive about a guy like Chris and how he operates. He is not 'jumping into the deep end' without first knowing a lot about what is in that water. He is not being hired as an intern, but as the guy who will led the team.

No argument. But I am saying he's no expert in the field. He's an expert in a different and fairly unrelated field. He has not spent years working on AI, driving, machine intelligence, or even realtime control software. Developer tools are essentially big, complex batch processing programs. It's totally different.

But he'll be running the operation and will have experienced experts working for him, people who have spent years trying (and failing) to produce autonomous driving systems. He'll be great at leading the effort. Just not an expert in the field.

 

[Nigel Tufnel]

And the winner for Best AutoPilot is: Tesla Motors AP2!
And here to accept the award is Elon Musk.
"I want to thank all of you for your support, I feel vindicated given all the negative comments about our latest AP version..."
Wait, I'm sorry Elon, apparently some TMC forum member switched the cards as a joke.

 

[Matias]

Elon, on the other hand, has said that autonomous driving is pretty much a "solved problem".

He also said, that it is possible to dig tunnels ten times faster than now, before he had ever seen a tunnel digging machine. He obviously says a lot of things..

If you read this article, this tells that Elon sees a tunnel digging machine for a first time. He made his claims before this.

Elon Musk Is Really Boring

 

[NerdUno]

For those entering this thread for the first time, let me share my decoder badge. After I read a post, I check the number of messages the author has posted previously because many don't acknowledge that they've only owned an AP1 vehicle. If the number is over 1,000, it tells me we're more likely than not dealing with an Elon Cheerleader. That's not a bad thing, but it should tell you that they've been enjoying a Tesla that was backed by 10-year-old Mobileye hardware and software that is no longer available in AP2 vehicles. All of the studies they cite and the growing pains they describe with AP1 have very little to do with AP2 because it's a fundamentally different technology with all-new hardware and completely different software AND NO TRACK RECORD. Aside from the fact that both systems are used to steer cars, they have almost nothing else in common.

My plea to the Cheerleader Squad would be to stop telling all of us that own AP2 vehicles about your growing pains with AP1 and how great and safe AP1 is today. It really doesn't matter, and it really isn't relevant. The fact that some group can become fluent in the German language in less than a year after having lived in Germany for a decade has zero relevance to whether a different group can master Chinese having never lived there and having never met a person that spoke Chinese.

 

[GSP]

He also said, that it is possible to dig tunnels ten times faster than now, before he had ever seen a tunnel digging machine. He obviously says a lot of things..

If you read this article, this tells that Elon sees a tunnel digging machine for a first time. He made his claims before this.

Elon Musk Is Really Boring

Great article about TBC, The Boring Company. Thanks for posting.

GSP

 

[Vitold]

The bike lane had nothing to do with it.

How is it that you do not understand? When AP1 first arrived (7.0) these situations were a piece of cake. Surface road was easy pickings for AP1. Now you need to take over every 10 seconds or you're dead. The point isn't that ap2 hasn't reached parity with ap1. Its that 17.7.2 for surface roads is dangerous. that's literally from every 17.7.2 video I have seen.

Again Ap2 is good on highway/freeway. No one is denying that. but its beyond dangerous on most surface roads.

If you need to take over every 10 seconds - you should not engage it as it's not safe. AP2 is able to drive on local roads but not all local roads. I'm baffled however why Tesla is allowing repeated engagement/disengagement - in such circumstances they should geo-lock such streets until new software revision.

 

[disagree]

I'm baffled however why Tesla is allowing repeated engagement/disengagement - in such circumstances they should geo-lock such streets until new software revision.

Agreed. It sure seems that shadow mode could determine safe areas and then these areas would be given green lights to offer autosteer. That is a pleasant order.

Or we could all risk safety and property damage trying out undercooked alpha software for Tesla. When we realize that we almost took out a row of parked cars or veered into oncoming traffic, we will have learned not to use it in a certain spot.

The latter sure seems like less work for Tesla, who gets to say "we released EAP software in 2016." Better to do that than honestly confront that EAP is a mess and was known months ago to be far, far away from realization.

 

[FluxCap]

HI,
This is my first time on this site, I've had my Model S for a month now. In the last couple days, I have tried the autosteer on the freeway, and a variety of roads. i think it never goes more than 10 seconds without doing something that is not just wrong, but potentially hazardous. The speed and acceleration of the car makes it particularly scary: this morning, it suddenly accelerated right toward a wall behind a sharp turn in front of me.
I am not just an owner, but a stockholder like many of you. Believe me, I do not want to find fault if I don't have to. But this software seems very far from becoming usable. When Chris Lattner, the new software engineer from Apple was hired a month ago, he described autonomous driving: "It is also a very, very hard technology problem." Notice the use of two 'very's' by someone who really knows what he is talking about.
I personally now believe that the separation from Mobileye was far more damaging than Elon will admit. Also, the loss of his earlier engineer, that hurt enough to compel Elon to sue him.
Although he is clearly a genius, and perhaps the man I most admire, Elon's confidence that they can quickly do their own autopilot looks like classic hubris. Hubris is the foundation of most tragic literature, and this step in Tesla's journey is starting to look like tragedy (waiting to happen). I will shocked (pleasantly) if anything usable occurs in the next three months, and possibly not this year. I even wonder about his claim that HW2 is capable of level 5. Audi supposedly has two radar sensors, not one. And Elon has already mentioned they will likely need to replace and upgrade the Nvidia processor. So a lot of warning flags have been thrown into the air....
I am certainly not 'angry' and enjoy driving, so I'm not as impatient as some on this thread. And great ambitions deserve some extra dispensation. But I do find this disappointing.

Thanks for joining TMC. Out of curiosity, what's your VIN, and are you long/short TSLA stock?

 

[electracity]

So it was released because pulling AP as a Tesla feature would kill sales and call into question the competency of the company at a critical time. Yikes.

I do agree with an earlier poster in this thread that it should be fixed by summer. Assuming Musk has managed to hold together most of the team that has worked on this stuff.

Some model 3 reservation holders are going to be disappointed when their new Tesla doesn't self-deliver to their house.

 

[Vitold]

I don't know of any reason to believe that Chris Lattner knows anything about autonomous driving software or AI. He was a compiler and developer tools guy at Apple. I'm sure he'll learn fast, and I'm also sure that Tesla has many people who are experts in the important areas, but he's not "someone who really knows". Or wasn't when he said that anyway.

Elon, on the other hand, has said that autonomous driving is pretty much a "solved problem".

Chris has extensive experience in managing and leading teams and was successful developing tools that increased performance and reliability.

Knowing AI is only one part - one that even few peoples can do in a Comma...ai - making sure that development is not introducing new bugs with every new code revision is another.

 

[Vitold]

HI,
This is my first time on this site, I've had my Model S for a month now. In the last couple days, I have tried the autosteer on the freeway, and a variety of roads. i think it never goes more than 10 seconds without doing something that is not just wrong, but potentially hazardous. The speed and acceleration of the car makes it particularly scary: this morning, it suddenly accelerated right toward a wall behind a sharp turn in front of me.
I am not just an owner, but a stockholder like many of you. Believe me, I do not want to find fault if I don't have to. But this software seems very far from becoming usable. When Chris Lattner, the new software engineer from Apple was hired a month ago, he described autonomous driving: "It is also a very, very hard technology problem." Notice the use of two 'very's' by someone who really knows what he is talking about.
I personally now believe that the separation from Mobileye was far more damaging than Elon will admit. Also, the loss of his earlier engineer, that hurt enough to compel Elon to sue him.
Although he is clearly a genius, and perhaps the man I most admire, Elon's confidence that they can quickly do their own autopilot looks like classic hubris. Hubris is the foundation of most tragic literature, and this step in Tesla's journey is starting to look like tragedy (waiting to happen). I will shocked (pleasantly) if anything usable occurs in the next three months, and possibly not this year. I even wonder about his claim that HW2 is capable of level 5. Audi supposedly has two radar sensors, not one. And Elon has already mentioned they will likely need to replace and upgrade the Nvidia processor. So a lot of warning flags have been thrown into the air....
I am certainly not 'angry' and enjoy driving, so I'm not as impatient as some on this thread. And great ambitions deserve some extra dispensation. But I do find this disappointing.

Have your car checked - you may have camera alignment issues. In any case, keep in mind that local road feature has just been released. If it does not work for your local roads - just wait for new releases - don't just engage it and hope for a different outcome ( like the saying goes, computers can help up make mistakes faster ).

Autonomous driving is a 'very hard technology problem' because of packaging (no one wants to drive around with trunk full of computers and roof rack full of sensors). Chris is a 'compiler guy', and great addition to Tesla to make sure that code works and performs.

It would be silly to assume that current AP2 is end of all considering huge advances in computing power and efficacy that occur every year. I'm guessing that in a year or two we may see custom chip developed by AMD per Tesla specifications. That is why EM said that current system is up-gradable.

 

[u00mem9]

This situation is about to spiral out of control.

For the last 5 months, Tesla has been able to sell cars based on the assumption that Autopilot functionality was eminent. I'm not sure, but I assume they were conducting test drives with older AP1 cars in inventory for the purpose of demonstrating what customers can expect (at minimum). That period is over.

Now, a potential customer will be (should be?) shown an AP2 vehicle with the current release. That customer will realize this is not currently a usable feature. That has to negatively affect sales.

Personally, my test drive and a subsequent rental via Toro made me realize AP1 was a reality and would change my quality of life that day! That sold a car. If I had experienced what I've seen in those videos above, I never would have made the purchase.

 

[Vitold]

This situation is about to spiral out of control.

For the last 5 months, Tesla has been able to sell cars based on the assumption that Autopilot functionality was eminent. I'm not sure, but I assume they were conducting test drives with older AP1 cars in inventory for the purpose of demonstrating what customers can expect (at minimum). That period is over.

Now, a potential customer will be (should be?) shown an AP2 vehicle with the current release. That customer will realize this is not currently a usable feature. That has to negatively affect sales.

Personally, my test drive and a subsequent rental via Toro made me realize AP1 was a reality and would change my quality of life that day! That sold a car. If I had experienced what I've seen in those videos above, I never would have made the purchase.

If you were here long enough you'd have known that everything is done right...but in Tesla time.

 

[brianman]

For the last 5 months, Tesla has been able to sell cars based on the assumption that Autopilot functionality was eminent. I'm not sure, but I assume they were conducting test drives with older AP1 cars in inventory for the purpose of demonstrating what customers can expect (at minimum). That period is over.

I had similar concerns regarding "here's how Autopilot works" demoed on AP1 vs. "here's how Autopilot works" on a delivered car ordered based on that experience when I saw these videos:
Tesla invites Europeans to take 'Drive To Believe' challenge: one week with Model S

 

[croman]

Local roads is far from perfect, but honestly, it performs just a bit worse than AP1. Not bad for a first iteration. AP1 isn't amazing in the areas where we're seeing complete failure by AP2. AP1 is less likely to veer into oncoming traffic and have the type of disaster moments I've seen from AP2 but its a lot more mature. I am willing to be patient in this regard. I just want Tesla to release safety features that are 100% and not half-baked. The convenience aspect is one thing, failure of safety features is a far different story and one that I am not willing to tolerate in the least (no one should have to).

 

[disagree]

Local roads is far from perfect, but honestly, it performs just a bit worse than AP1. Not bad for a first iteration. AP1 isn't amazing in the areas where we're seeing complete failure by AP2. AP1 is less likely to veer into oncoming traffic and have the type of disaster moments I've seen from AP2 but its a lot more mature. I am willing to be patient in this regard. I just want Tesla to release safety features that are 100% and not half-baked. The convenience aspect is one thing, failure of safety features is a far different story and one that I am not willing to tolerate in the least (no one should have to).

The comparison between how AP1 and EAP handle local roads is really just a curiosity (maybe even a red herring -- not saying you're using it for one, but others seem swept up with it). AP1 was never -- to my knowledge -- given a Local Road Autosteer update. Is this right?

On its own "merits", this EAP local road release is not safe and shouldn't have been sent out. It doesn't get EAP any closer to parity with AP1, the immediate pressing goal (basic features like smart wipers, summon, etc. and highway speed autosteer). It is a mind-boggling decision to rush it out.

 

[drshan]

I've seen nothing indicating this early AP2 is worse than the similarly early AP1. And even the more fully baked AP1 now still does not navigate intersections or roads without lane markings or traffic circles very well or at all.

AP is best for well marked divided highways and stop and go traffic surrounded by slow moving cars. And both of those situations are when driving is most tedious and thus when AP is most desirable.

I think Tesla has mismanaged the communication on what to expect from EAP as it progresses to AP1 parity.

Since they acknowledge that AP2 isn't at AP1 parity right now, video showing AP2 not handling situations that AP1 can't even handle isn't a fail. That is just what it means to not yet have surpassed AP1 abilities yet.

Really? This is me on a local road yesterday with the latest release 17.7.2.