HW2.5 capabilities

[TheTron]

There are two things I think we can say with some certainty.

1) Tesla will not have the resources to drive two separate FSD development programs, one for AP2 and another for 2.5. Getting FSD working on one platform is a gargantuan challenge as it is.

2) Tesla cannot stop developing AP2 FSD until they are at least in a position where they have a working solution, albeit one that fails regulatory approval. Ditching AP2 before this point would lead to a *sugar* storm of bad PR and litigation.

These points lead me to believe we will either see a computer retrofit for AP2 FSD cars to allow them to run the true FSD firmware designed for AP2.5. These retrofitted cars would not have the redundant wiring of the new cars and so might fail regulatory approval at some point in the future.

It is also possible that, while the two platforms now have different firmwares, these firmwares are compiled from almost identical source code and share the same neural nets. The 2.5 version would have some additional features like the cabin camera driver monitoring but the basic FSD functionality would be shared between the two platforms.

I'm hoping the second option comes to pass. My worry with the retrofit is that it may not happen until the end of the development of FSD meaning us AP2 owners would miss out on the incremental release of FSD features over the coming months (or years?!)

 

[lunitiks]

The way I see it, 2.5 is "mje".

Arguable. I'm not sure what it means.

Either way, still no "Enhanced" Autopilot. Really, wtf

 

[FlatSix911]

There are two things I think we can say with some certainty.

1) Tesla will not have the resources to drive two separate FSD development programs, one for AP2 and another for 2.5. Getting FSD working on one platform is a gargantuan challenge as it is.

2) Tesla cannot stop developing AP2 FSD until they are at least in a position where they have a working solution, albeit one that fails regulatory approval. Ditching AP2 before this point would lead to a *sugar* storm of bad PR and litigation.

These points lead me to believe we will either see a computer retrofit for AP2 FSD cars to allow them to run the true FSD firmware designed for AP2.5. These retrofitted cars would not have the redundant wiring of the new cars and so might fail regulatory approval at some point in the future. It is also possible that, while the two platforms now have different firmwares, these firmwares are compiled from almost identical source code and share the same neural nets. The 2.5 version would have some additional features like the cabin camera driver monitoring but the basic FSD functionality would be shared between the two platforms. I'm hoping the second option comes to pass. My worry with the retrofit is that it may not happen until the end of the development of FSD meaning us AP2 owners would miss out on the incremental release of FSD features over the coming months (or years?!)

I think a retrofit of AP2 is highly doubtful ... just as likely as a retrofit of AP1. Tesla future resources will go to AP2.5 and 3.0

 

[ab26]

I think a retrofit of AP2 is highly doubtful ... just as likely as a retrofit of AP1. Tesla future resources will go to AP2.5 and 3.0

When you quote AP3.0, are you stating the obvious that AP3.0 will happen sometime in the future, or is this based on any indication that there's already AP3.0 (which is different from AP2.5)?

 

[FlatSix911]

When you quote AP3.0, are you stating the obvious that AP3.0 will happen sometime in the future, or is this based on any indication that there's already AP3.0 (which is different from AP2.5)?

The engineers in Palo Alto are already working on 3.0 ... they will not say any more than that.

 

[chillaban]

The engineers in Palo Alto are already working on 3.0 ... they will not say any more than that.

I am torn between the two sides of this....

As a Silicon Valley engineer, I have to say, they would be doing a great disservice to themselves if they weren't working on AP 2.5 and now AP 3.0. The underlying technology improves so fast year-over-year, not to mention the way popular approaches evolve over a slightly longer timeframe. They'd be setting themselves up to be eclipsed by newer competitors if they didn't start working on the next generation as the current is being announced/shipped.

With that said, as a Tesla owner, I still expect them to deliver on the originally promised AP2.0 feature set, especially given that we are getting close to the anniversary of its announcement and the best we could say is that highway driving is close-ish to AP1 but not a trace of "enhanced" features exists yet.

As far as 2.0 vs 2.5, I also am on the side of believing it to be primarily for redundancy and should not represent a major split in functionality between AP2.0 and AP2.5 at least in terms of the immediate functionality we've been lacking (e.g. robust L2 autonomy especially in city conditions, transitioning into L3).

 

[mandersen]

So when does the AP2.5 hardware start shipping? With an order in and a month+ out on delivery trying to make decisions on what I want to do with my order for a car I will own for the next 3 years or so.

 

[verygreen]

So when does the AP2.5 hardware start shipping? With an order in and a month+ out on delivery trying to make decisions on what I want to do with my order for a car I will own for the next 3 years or so.

If you are buying the car for FSD or APE - don't buy now, wait until it's actually released.
I guess there should be some sort of a FAQ item at this point.

 

[bswn1]

If you are buying the car for FSD or APE - don't buy now, wait until it's actually released.
I guess there should be some sort of a FAQ item at this point.

What does the atlas connections imply? Is the new map roll-out imminent?

Also, do we know what the NN is actually doing at this stage?

 

[mandersen]

If you are buying the car for FSD or APE - don't buy now, wait until it's actually released.
I guess there should be some sort of a FAQ item at this point.

More concerned with dropping $100k for a car and having new hardware show up 30 days later.

 

[mitchellh3]

More concerned with dropping $100k for a car and having new hardware show up 30 days later.

As someone who has owned two Model S': no matter when you purchase the car there are going to be changes in the next quarter or two.

As @verygreen said, if FSD or EAP are a core requirement for you and you'd be regretful immediately if hardware changed to allow that more easily on another car, you should not buy this car now. There are going to be minor hardware upgrades constantly. And probably a major one every 2 to 3 years.

You've gotta do it because you love the car you're buying now, not the car that it might be later.

 

[chillaban]

As someone who has owned two Model S': no matter when you purchase the car there are going to be changes in the next quarter or two.

As @verygreen said, if FSD or EAP are a core requirement for you and you'd be regretful immediately if hardware changed to allow that more easily on another car, you should not buy this car now. There are going to be minor hardware upgrades constantly. And probably a major one every 2 to 3 years.

You've gotta do it because you love the car you're buying now, not the car that it might be later.

Totally agreed with your message. I would just add that for me, I'd say if "E"AP in its current state is not what interests you (e.g. you're looking towards the yet-to-be-delivered features), you'll probably feel disappointed within quarters of taking delivery.

Ever since the 17.26 series of software updates, I've found that EAP's Autosteer is already quite competent for my road trips and it's something I value a lot in my car. Yes there is a lot to be desired still, but there are going to be people who are happy with what EAP offers today and what it looks like it'll be able to offer in the near future.

But if you're structuring your purchase as an investment into the FSD capabilities, I think @verygreen and @mitchellh3 are right on the money: It's best to hold off, because it's more than likely that Tesla will have to iterate on the hardware and software multiple times before they know what it really takes to get there. And while they might retrofit early adopters or otherwise attempt to deliver on their original promises, it's more than likely the new cars rolling off the line at that time in the future will be in a better position to get to FSD.

 

[alcibiades]

More concerned with dropping $100k for a car and having new hardware show up 30 days later.

I'd wait, too. I bought my X right after the October 2016 announcement. I've now had my car since last December and not only do I not have the features of AP1, I also neither have any enhanced features and it seems that the announcement that all cars had the hardware for FSD was duplicitous at best.

If FSD matters to you, you'd do best to wait.

 

[FlatSix911]

So when does the AP2.5 hardware start shipping? With an order in and a month+ out on delivery trying to make decisions on what I want to do with my order for a car I will own for the next 3 years or so.

AP2.5 started shipping last month with the Model 3

 

[mandersen]

AP2.5 started shipping last month with the Model 3

What about the Model S?

 

[verygreen]

What about the Model S?

If you order now you'll get AP2.5, that does not mean they won't announce AP3 next month (to make it clear, does not mean they would either - I just don't know).

Some people think one really needs to wait until next year to make an order because more exciting stuff will come.

 

[stopcrazypp]

There's a redundancy requirement for L5. To me that explains the extra wiring and any other electronic backups.

If someone wants to chase down the dates of the NHTSA's L5 guidelines vs. AP 2.0 release, then we may find that AP 2.0 preceded the new NHTSA guidelines. Not sure, but that would explain a lot of these moves.

My take is that AP 2.5 is about ride-sharing and meeting this nebulous redundancy "requirement". I wrote a member article to that effect a couple of weeks ago.

Given the new information about an upgraded radar do we know whether or not the old radar supported double CAN-bus signaling? If it didn't, then this move could again be explained under the "redundancy" clause, and it's upgraded specs might have just been a bonus.

Is redundancy actually a requirement by the NHTSA guidelines? Any references?

If so, this certainly falls into a regulatory requirement. SAE Level 5 as defined does not require any redundancy at all.

So this may be a situation where despite all the people poo-pooing Tesla's caveats about regulatory requirements, it's actually something that is significant (could be a situation where Tesla gets to something that is L5, but is not legal because it lacks redundancy).

However, I would still caution against making too many assumptions. It seems everyone here is assuming all changes being made are for FSD when many of them may be unrelated.

 

[Alketi]

Is redundancy actually a requirement by the NHTSA guidelines? Any references?

If so, this certainly falls into a regulatory requirement. SAE Level 5 as defined does not require any redundancy at all.

Redundancy mentioned by NHTSA:

The process should describe design redundancies and safety strategies for handling cases of HAV system malfunctions.

Federal Automated Vehicles Policy - Sept. 2016

 

[scottf200]

Is redundancy actually a requirement by the NHTSA guidelines? Any references?

If so, this certainly falls into a regulatory requirement. SAE Level 5 as defined does not require any redundancy at all.

So this may be a situation where despite all the people poo-pooing Tesla's caveat's about regulatory requirements, it's actually something that is significant (could be a situation where Tesla gets to something that is L5, but is not legal because it lacks redundancy).

However, I would still caution against making too many assumptions. It seems everyone here is assuming all changes being made are for FSD when many of them may be unrelated.

Not sure if documented guidelines yet but articles about it...
Autonomous future will require new product development approaches - SAE International

System redundancy is vital

Vehicle system redundancy will be a key fixture of autonomous vehicle development, according to Jeremy Carlson, Senior Analyst, Autonomous Drivingfor IHS Automotive. “There’s already a pretty high bar of quality and reliability that goes into any engineered part of the vehicle,” Carlson told Automotive Engineering. “One of the big things that will change, though, is the need for redundant systems.”

That emphasis on redundancy will be akin to practices in the aviation industry, he noted. “There’s no consensus on what level of redundancy is needed, but we’re already seeing multiple sensors, such as camera, radar, and Lidar, for redundant input,” said Carlson, “And the need for redundant control will increase.”

Obstacle detection amid 360º views of the vehicle’s surroundings are an automated-car must, especially for those with SAE Level 5 (full) autonomy. For engineers, "it’s not just about figuring out what’s the graceful degradation from one level to the next, or from one system to a back-up system,” said Carlson, “It’s also about putting that technology into a vehicle where packaging space is always at a premium.”

According to industry experts, autonomous vehicles with human fallback are only a few years from being in the market, while autonomous vehicles with system-level fallback—known as SAE Levels 4 and 5 (high and full autonomy) could reach the U.S. market in the 2020-2030 timeframe.

 

[AnxietyRanger]

For those asking: the redundant wiring, power, performance and new radar of HW 2.5 should now be shipping in new Model S/X orders.

However we believe the interior camera may only ship on Model 3 at this time. It is expeced that is part of the self-driving suite going forward.

All of this may face major changes in the following quaters as well IMO as Model 3 ramps up and its next-gen developments further trickle to Model S/X.