The following is speculation based on the author’s experience, the known Model 3 feature set, and Tesla’s plans and recent statements.
Summary
The Model 3 reportedly contains Autopilot hardware dubbed “v2.5“
News outlets have yet to fully explore the meaning and purpose
Tesla isn’t providing a full explanation because they don’t want to tip their hand and they don’t need you to reserve a Model 3 — yet
Welcome to anti-sell and production-hell!
In a highly anticipated but somewhat awkward Model 3 delivery event, Elon Musk, ever the promoter, showed just one positive video of the car: the Model 3 being tested for side-impact where it effectively bounced off the testing device while one of the safest cars in the world, the Volvo S60, wrapped around it like a burrito.
Then he welcomed his fellow employees to production hell, showed a single slide that highlighted a handful of Model 3 specs, said nothing about the interior or any other features, and exited the stage faster than Jack Black hitting an empty mosh pit.
Don’t be fooled, this was the anti-sell.
Musk has been terrified of the Osborne effect since the original Model 3 unveiling, often publicly lamenting that people are mistaking the “3” for third generation and then taking that to mean it contains newer/better technology than S/X.
Note: It actually does have newer/better/mind-blowing technology, but not in the way most people were thinking. The technology in question will be the subject of this post.
In fact, so pervasive is Tesla’s concern for the Osborne effect that they still anti-sell the Model 3 on it’s own webpage. Scroll half-way down and they’ll happily tell you all the ways the Model S is superior. Don’t reserve a Model 3; buy a Model S today!
Despite this concerted anti-sell and after only two 30 minute presentations and a handful of tweets, the Model 3 reservation count exceeds 455,000, more than the entire 2016 worldwide sales of the BMW 3 series.
Oops.
OK, great. So Tesla fooled 450,000 people into thinking they were buying tickets to a rock concert. They’re all just going to cancel anyway. What’s Model 3’s big secret?
Let’s walk through this mystery, step-by-step.
The Great Autopilot Head-Fake
Reports have said that the Model 3 has, what some are calling, Autopilot HW 2.5.
Here’s how Tesla responded to this report:
“The internal name HW 2.5 is an overstatement, and instead it should be called something more like HW 2.1. This hardware set has some added computing and wiring redundancy, which very slightly improves reliability, but it does not have an additional Pascal GPU.”
(The part in bold is key and will make sense by the end of this article.)
Still, HW 2.5 is being interpreted to mean that Tesla has already obsoleted the autopilot hardware found in the S and X and/or realized it’s not sufficient for full self-driving.
For example, the Verge wrote:
The updated onboard computer, dubbed “HW 2.5,” would seem to contradict the previous vow by Tesla CEO Elon Musk that all vehicles released since October 2016 would have the hardware necessary to achieve “full self-driving capabilities.
On Twitter, some of Tesla’s biggest fans have reached similar conclusions:
Are they right? Did Tesla just declare AP hardware v2.0 incapable of full self-driving?
Well, as always, there’s another side to the story. And — oh what a side it is.
First, let’s recall Tesla’s public plans.
Tesla’s Future Ride Sharing Network
In the second installment of his Master Plan, Musk outlined the future Tesla Ride Sharing Network as follows:
When true self-driving is approved by regulators, it will mean that you will be able to summon your Tesla from pretty much anywhere. Once it picks you up, you will be able to sleep, read or do anything else enroute to your destination.
You will also be able to add your car to the Tesla shared fleet just by tapping a button on the Tesla phone app and have it generate income for you while you’re at work or on vacation, significantly offsetting and at times potentially exceeding the monthly loan or lease cost.
Now let’s peel the layers of the HW 2.5 mystery.
There Can Be Only One
The Model 3, as it turns out, is Tesla’s Highlander.
It’s the only car in Tesla’s current lineup capable of someday joining its ride sharing network. The S and X, despite appearing to share the same autopilot hardware, will not make it in current form, nor do they even make sense as robotaxis, given this imminent and cheaper alternative.
Aha! So everyone was right!
Let’s note the important distinction that everyone seems to be missing:
Full self driving does not necessarily equate to ride sharing
Tesla is not saying that the S/X hardware can’t achieve full self-driving.
And yet they’re not explaining the Model 3’s distinctive features because they’re already drowning in reservations and would rather you buy an S or an X.
But they left too many clues.
The Model 3 brings forward two very important innovations, one interior, one exterior. And, HW 2.5 is what’s needed to tie them together.
About That Model 3 Interior
The Model 3’s interior is certainly striking, if only for its minimalism. A quick comparison with two of its competitors.
Model 3 interior
2018 Mercedes C-Class interior
2018 BMW 3-Series interior
From this, we can surmise that the Model 3 interior serves three purposes simultaneously.
It’s strikingly unique and unapologetically futuristic
It’s easier to manufacture (fewer parts)
It’s built for autonomous ride sharing
Yes, that last one. The Model 3 is built from the ground up to be the first candidate in Tesla’s future ride sharing network and the interior is the first piece of the puzzle.
The unique 15-inch screen is a hugely important feature.
Save for the window actuators (which are required) there are no other physical controls.
No radio.
No trunk latch.
No glove box handle.
No fiddly heating/air-conditioning vents.
Software Controls Go to the Next Level and Beyond
Who cares? Tesla’s just cheap, right?
Not just cheap…
Not having physical controls means everything in the Model 3 is software controlled. This means everything is configurable and remembered. Everything. And more importantly, it’s remembered per person.
This also makes the main display the access point in order to do things like open the trunk or the glove box, or, in general, to operate anything.
Access to configurable touch screen controls can be either prevented or granted.
What are the first two concerns people mention when asked if they’d hypothetically share their car with others in a ride-sharing network?
What if I leave something valuable in the car?
What if someone pukes/destroys my interior?
As just illustrated, Tesla addressed the first issue through 100% software controls. Permissions for anything can be granted to the owner, but restricted from other parties.
The glove box now acts as a safe that you don’t need to remember to lock.
The front and rear trunks are inaccessible without permission.
Nothing proprietary or personal can be accessed from the main display
This has never been done before in the history of automotive. And we’re only one third of the way through the mystery.
What about the second issue?
Monitoring the Interior
At CES 207 nVidia announced, among other things, AI Co-Pilot.
They listed its features as facial recognition, head position and gaze tracking, natural language processing, voice speech capabilities, augmented lip reading, external environmental awareness.
As one would guess, it features a camera to keep an eye on the driver and interior cabin.
The what? What was that second one?
The interior cabin. It monitors the interior cabin.
This would address the second piece of the ride-sharing interior puzzle: what if someone damages the interior?
And, in pouring over a video of the Model 3, someone noticed this.
An internal camera.
So, Tesla is not only using nVidia’s Drive PX2 board for autopilot, but they now appear to be using something very similar to its Co-Pilot technology to monitor the interior.
Let’s restate that for clarity — the Model 3 has a camera that can monitor the interior cabin in the event a third party damages something.
(Interior cameras will undoubtedly be standard in future robotaxi’s. And, yes, the car will likely not operate if the camera view is blocked. So keep your masking tape and magic markers at home.)
Also note that such a camera, not present on the S/X, would require computing power and new wiring to the Drive PX2/Co-Pilot board. Let’s recall Tesla’s statement to Electrek:
This hardware set has some added computing and wiring redundancy, which very slightly improves reliability, but it does not have an additional Pascal GPU.”
Are we getting it? Tesla needed new hardware to address two of the biggest concerns in sharing one’s vehicle with a third party: restricting access to valuables and features, and monitoring the interior.
But all of this still doesn’t work without one more innovation.
The Final Piece of the Ride-Sharing Puzzle
First we inserted keys.
Then we inserted key fobs.
Now we have wireless key fobs that we keep in our pockets and can start the car with the push of a button.
Yet, none of these interactions are sufficient for a future ride sharing environment.
You can’t lend your key fob to everyone who would ride in your car. But, that’s OK. It’s about to become obsolete.
The Backup Entry Option
As Model 3 videos began to surface it was quickly discovered that Tesla is replacing the key fob with an NFC key card. Entry via key card involves swiping it against the B-pillar, and then placing it in the cup holder.
While interesting, and simultaneously futuristic and (seemingly) a step backward, this is a red herring.
The key card is the secondary form of entry, and it’s purpose is to obsolete the key fob, not replace it. There’s something much more important in play.
The REAL Entry Option
The main form of entry to the Model 3 is actually your cell phone.
With the Tesla app and an Internet connection, one can press a button and lock/unlock the doors from anywhere in the world. Both the S and X also work this way.
Wait. What if you don’t have an Internet connection?
And how is unlocking your phone, launching an app, waiting for it to load, and pressing a button possibly better than a key fob in your pocket??
Almost there.
Enter The Magic Protocol
As we know, everything’s better with Bluetooth. And so is ride-sharing.
Starting with the Model 3, Tesla’s mobile app will also run a persistent background service that uses Bluetooth LE to directly connect to the vehicle.
Bluetooth LE is an advertising/scanning-based approach that one can use to securely transmit encrypted public/private keys and data. It does not require explicit pairing like traditional Bluetooth.
In other words, with your phone in your pocket, the car will not only unlock as you approach but also know who you are and instantly configure itself with your preferences, privileges, and restrictions.
And that my friends is the final piece of the puzzle.
Tesla’s Entire Plan Revealed
Let’s recap. Tesla needed new hardware to enable new features, not to change self-driving performance.
As such, Tesla is about to deliver, right under everyone’s noses:
Autopilot hardware it believes will be capable of full autonomy.
A 100% software-controlled vehicle that’s accessed identically by either its owner or a random person, and is fully customizable for each.
A camera to monitor the interior.
Vehicle entry with a mobile phone.
Autopilot hardware. A fully software controlled vehicle. An interior camera. Entry via a cell phone.
Autopilot hardware. A fully software controlled vehicle. An interior camera. Entry via a cell phone.
Are we getting it yet?
The Model 3’s design is NOT just to save money.
HW 2.5 isn’t about self-driving.
It’s about ride-sharing.
Conclusions
Despite misguided criticism, Tesla has not obsoleted Autopilot 2.0 hardware and this isn’t a “backup plan.” AP HW 2.5 is not about self-driving; it’s about ride-sharing. And Tesla doesn’t want to explain it — yet.
In short, with the Model 3, Tesla is delivering a mind-blowing electric vehicle that:
Incorporates everything Tesla believes is necessary for full autonomy
Incorporates everything needed to join its future ride sharing network
Will recognize you as you approach
Unlock your door and adjust your seat position
Restrict any parts of the car you’re not approved to access
Set the car to your temperature and aim the air vents where you prefer
Set all your radio presets
Queue your favorite streaming music or podcasts
(Someday) play your favorite movies and television shows
And, the “you” in question can be a complete stranger.
And all this happens instantly as you approach, without any manual interaction.
Expect more discussion on this from tech and auto industry reporters, and eventually the general public.
This is what Apple, GM, and others aspire to deliver three or more years from now.
Tesla isn’t going to explain it because, for now, they’d rather you buy an S or an X.
Despite keeping this secret they still have over 455,000 reservations. And counting.
The bottom line is that contrary to what Tesla says, the Model 3 is the most technologically advanced car that Tesla or anyone else has ever built.
And it’s not even close.
Tesla is innovating at a rate unmatched in the auto industry. Not only are they planning years ahead but they’re also executing in a way that none of their peers can even conceive.
For contrast:
Mazda just announced a 20% improvement breakthrough in an ICE engine that it hopes to bring to market in 2019.
Ford is adding 25HP to this year’s Mustang, and also adding a mute button, so the engine doesn’t annoy your neighbors.
These are ocean liners trying to catch a catamaran and they’re not even in the right body of water.
Is it any wonder Tesla accelerated their bond sale when it could have waited until 2018?
TMC Member Alketi is an electrical/software engineer with 25 years experience. He is also a Model 3 reservation holder.
Summary
The Model 3 reportedly contains Autopilot hardware dubbed “v2.5“
News outlets have yet to fully explore the meaning and purpose
Tesla isn’t providing a full explanation because they don’t want to tip their hand and they don’t need you to reserve a Model 3 — yet
Welcome to anti-sell and production-hell!
In a highly anticipated but somewhat awkward Model 3 delivery event, Elon Musk, ever the promoter, showed just one positive video of the car: the Model 3 being tested for side-impact where it effectively bounced off the testing device while one of the safest cars in the world, the Volvo S60, wrapped around it like a burrito.
Then he welcomed his fellow employees to production hell, showed a single slide that highlighted a handful of Model 3 specs, said nothing about the interior or any other features, and exited the stage faster than Jack Black hitting an empty mosh pit.
Don’t be fooled, this was the anti-sell.
Musk has been terrified of the Osborne effect since the original Model 3 unveiling, often publicly lamenting that people are mistaking the “3” for third generation and then taking that to mean it contains newer/better technology than S/X.
Note: It actually does have newer/better/mind-blowing technology, but not in the way most people were thinking. The technology in question will be the subject of this post.
In fact, so pervasive is Tesla’s concern for the Osborne effect that they still anti-sell the Model 3 on it’s own webpage. Scroll half-way down and they’ll happily tell you all the ways the Model S is superior. Don’t reserve a Model 3; buy a Model S today!
Despite this concerted anti-sell and after only two 30 minute presentations and a handful of tweets, the Model 3 reservation count exceeds 455,000, more than the entire 2016 worldwide sales of the BMW 3 series.
Oops.
OK, great. So Tesla fooled 450,000 people into thinking they were buying tickets to a rock concert. They’re all just going to cancel anyway. What’s Model 3’s big secret?
Let’s walk through this mystery, step-by-step.
The Great Autopilot Head-Fake
Reports have said that the Model 3 has, what some are calling, Autopilot HW 2.5.
Here’s how Tesla responded to this report:
“The internal name HW 2.5 is an overstatement, and instead it should be called something more like HW 2.1. This hardware set has some added computing and wiring redundancy, which very slightly improves reliability, but it does not have an additional Pascal GPU.”
(The part in bold is key and will make sense by the end of this article.)
Still, HW 2.5 is being interpreted to mean that Tesla has already obsoleted the autopilot hardware found in the S and X and/or realized it’s not sufficient for full self-driving.
For example, the Verge wrote:
The updated onboard computer, dubbed “HW 2.5,” would seem to contradict the previous vow by Tesla CEO Elon Musk that all vehicles released since October 2016 would have the hardware necessary to achieve “full self-driving capabilities.
On Twitter, some of Tesla’s biggest fans have reached similar conclusions:
Well, as always, there’s another side to the story. And — oh what a side it is.
First, let’s recall Tesla’s public plans.
Tesla’s Future Ride Sharing Network
In the second installment of his Master Plan, Musk outlined the future Tesla Ride Sharing Network as follows:
When true self-driving is approved by regulators, it will mean that you will be able to summon your Tesla from pretty much anywhere. Once it picks you up, you will be able to sleep, read or do anything else enroute to your destination.
You will also be able to add your car to the Tesla shared fleet just by tapping a button on the Tesla phone app and have it generate income for you while you’re at work or on vacation, significantly offsetting and at times potentially exceeding the monthly loan or lease cost.
Now let’s peel the layers of the HW 2.5 mystery.
There Can Be Only One
The Model 3, as it turns out, is Tesla’s Highlander.
It’s the only car in Tesla’s current lineup capable of someday joining its ride sharing network. The S and X, despite appearing to share the same autopilot hardware, will not make it in current form, nor do they even make sense as robotaxis, given this imminent and cheaper alternative.
Aha! So everyone was right!
Let’s note the important distinction that everyone seems to be missing:
Full self driving does not necessarily equate to ride sharing
Tesla is not saying that the S/X hardware can’t achieve full self-driving.
And yet they’re not explaining the Model 3’s distinctive features because they’re already drowning in reservations and would rather you buy an S or an X.
But they left too many clues.
The Model 3 brings forward two very important innovations, one interior, one exterior. And, HW 2.5 is what’s needed to tie them together.
About That Model 3 Interior
The Model 3’s interior is certainly striking, if only for its minimalism. A quick comparison with two of its competitors.
From this, we can surmise that the Model 3 interior serves three purposes simultaneously.
It’s strikingly unique and unapologetically futuristic
It’s easier to manufacture (fewer parts)
It’s built for autonomous ride sharing
Yes, that last one. The Model 3 is built from the ground up to be the first candidate in Tesla’s future ride sharing network and the interior is the first piece of the puzzle.
The unique 15-inch screen is a hugely important feature.
Save for the window actuators (which are required) there are no other physical controls.
No radio.
No trunk latch.
No glove box handle.
No fiddly heating/air-conditioning vents.
Software Controls Go to the Next Level and Beyond
Who cares? Tesla’s just cheap, right?
Not just cheap…
Not having physical controls means everything in the Model 3 is software controlled. This means everything is configurable and remembered. Everything. And more importantly, it’s remembered per person.
This also makes the main display the access point in order to do things like open the trunk or the glove box, or, in general, to operate anything.
Access to configurable touch screen controls can be either prevented or granted.
What are the first two concerns people mention when asked if they’d hypothetically share their car with others in a ride-sharing network?
What if I leave something valuable in the car?
What if someone pukes/destroys my interior?
As just illustrated, Tesla addressed the first issue through 100% software controls. Permissions for anything can be granted to the owner, but restricted from other parties.
The glove box now acts as a safe that you don’t need to remember to lock.
The front and rear trunks are inaccessible without permission.
Nothing proprietary or personal can be accessed from the main display
This has never been done before in the history of automotive. And we’re only one third of the way through the mystery.
What about the second issue?
Monitoring the Interior
At CES 207 nVidia announced, among other things, AI Co-Pilot.
They listed its features as facial recognition, head position and gaze tracking, natural language processing, voice speech capabilities, augmented lip reading, external environmental awareness.
As one would guess, it features a camera to keep an eye on the driver and interior cabin.
The what? What was that second one?
The interior cabin. It monitors the interior cabin.
This would address the second piece of the ride-sharing interior puzzle: what if someone damages the interior?
And, in pouring over a video of the Model 3, someone noticed this.
An internal camera.
So, Tesla is not only using nVidia’s Drive PX2 board for autopilot, but they now appear to be using something very similar to its Co-Pilot technology to monitor the interior.
Let’s restate that for clarity — the Model 3 has a camera that can monitor the interior cabin in the event a third party damages something.
(Interior cameras will undoubtedly be standard in future robotaxi’s. And, yes, the car will likely not operate if the camera view is blocked. So keep your masking tape and magic markers at home.)
Also note that such a camera, not present on the S/X, would require computing power and new wiring to the Drive PX2/Co-Pilot board. Let’s recall Tesla’s statement to Electrek:
This hardware set has some added computing and wiring redundancy, which very slightly improves reliability, but it does not have an additional Pascal GPU.”
Are we getting it? Tesla needed new hardware to address two of the biggest concerns in sharing one’s vehicle with a third party: restricting access to valuables and features, and monitoring the interior.
But all of this still doesn’t work without one more innovation.
The Final Piece of the Ride-Sharing Puzzle
Then we inserted key fobs.
Now we have wireless key fobs that we keep in our pockets and can start the car with the push of a button.
Yet, none of these interactions are sufficient for a future ride sharing environment.
You can’t lend your key fob to everyone who would ride in your car. But, that’s OK. It’s about to become obsolete.
The Backup Entry Option
As Model 3 videos began to surface it was quickly discovered that Tesla is replacing the key fob with an NFC key card. Entry via key card involves swiping it against the B-pillar, and then placing it in the cup holder.
The key card is the secondary form of entry, and it’s purpose is to obsolete the key fob, not replace it. There’s something much more important in play.
The REAL Entry Option
With the Tesla app and an Internet connection, one can press a button and lock/unlock the doors from anywhere in the world. Both the S and X also work this way.
Wait. What if you don’t have an Internet connection?
And how is unlocking your phone, launching an app, waiting for it to load, and pressing a button possibly better than a key fob in your pocket??
Almost there.
Enter The Magic Protocol
As we know, everything’s better with Bluetooth. And so is ride-sharing.
Starting with the Model 3, Tesla’s mobile app will also run a persistent background service that uses Bluetooth LE to directly connect to the vehicle.
Bluetooth LE is an advertising/scanning-based approach that one can use to securely transmit encrypted public/private keys and data. It does not require explicit pairing like traditional Bluetooth.
In other words, with your phone in your pocket, the car will not only unlock as you approach but also know who you are and instantly configure itself with your preferences, privileges, and restrictions.
And that my friends is the final piece of the puzzle.
Tesla’s Entire Plan Revealed
Let’s recap. Tesla needed new hardware to enable new features, not to change self-driving performance.
As such, Tesla is about to deliver, right under everyone’s noses:
Autopilot hardware it believes will be capable of full autonomy.
A 100% software-controlled vehicle that’s accessed identically by either its owner or a random person, and is fully customizable for each.
A camera to monitor the interior.
Vehicle entry with a mobile phone.
Autopilot hardware. A fully software controlled vehicle. An interior camera. Entry via a cell phone.
Autopilot hardware. A fully software controlled vehicle. An interior camera. Entry via a cell phone.
Are we getting it yet?
The Model 3’s design is NOT just to save money.
HW 2.5 isn’t about self-driving.
It’s about ride-sharing.
Conclusions
Despite misguided criticism, Tesla has not obsoleted Autopilot 2.0 hardware and this isn’t a “backup plan.” AP HW 2.5 is not about self-driving; it’s about ride-sharing. And Tesla doesn’t want to explain it — yet.
In short, with the Model 3, Tesla is delivering a mind-blowing electric vehicle that:
Incorporates everything Tesla believes is necessary for full autonomy
Incorporates everything needed to join its future ride sharing network
Will recognize you as you approach
Unlock your door and adjust your seat position
Restrict any parts of the car you’re not approved to access
Set the car to your temperature and aim the air vents where you prefer
Set all your radio presets
Queue your favorite streaming music or podcasts
(Someday) play your favorite movies and television shows
And, the “you” in question can be a complete stranger.
And all this happens instantly as you approach, without any manual interaction.
Expect more discussion on this from tech and auto industry reporters, and eventually the general public.
This is what Apple, GM, and others aspire to deliver three or more years from now.
Tesla isn’t going to explain it because, for now, they’d rather you buy an S or an X.
Despite keeping this secret they still have over 455,000 reservations. And counting.
The bottom line is that contrary to what Tesla says, the Model 3 is the most technologically advanced car that Tesla or anyone else has ever built.
And it’s not even close.
Tesla is innovating at a rate unmatched in the auto industry. Not only are they planning years ahead but they’re also executing in a way that none of their peers can even conceive.
For contrast:
Mazda just announced a 20% improvement breakthrough in an ICE engine that it hopes to bring to market in 2019.
Ford is adding 25HP to this year’s Mustang, and also adding a mute button, so the engine doesn’t annoy your neighbors.
These are ocean liners trying to catch a catamaran and they’re not even in the right body of water.
Is it any wonder Tesla accelerated their bond sale when it could have waited until 2018?
TMC Member Alketi is an electrical/software engineer with 25 years experience. He is also a Model 3 reservation holder.
Last edited by a moderator:
