Elon: Apps and Google Chrome in the Model S by late-2014

[Newscutter]

I'm a little tech-stupid so, break it down for me:

Are these things that could be updated and used on the EXISTING car or would this require a different head unit? (i.e. retrofit being either impossible or expensive)

 

[wcalvin]

Window on 17" for remote control of iphone/Androids.

 

[phat78boy]

If they could integrate Google Now, it would be the perfect car system to me. Sync my tiles from my phone/computer and allow me to use its voice commands and search functions...perfect. Chrome has this ability in a PC, sure hope they retain that for the Model S.

 

[richkae]

There might only be an installed base of 25k users, but who says native Tesla apps would have the Android pricepoint? If the apps cost $25 or $50 instead of $1, would that change a developers thinking?

Based on what I have observed from owners on this Board, I consider it likely that some of us would pay those amounts for some well-executed native apps (like Spotify or Waze).

250+ million android devices vs 25,000 Teslas...
If you made a good app for the Tesla you would probably be ecstatic to sell 100 copies.

The motivation for native Tesla apps is going to have to be helping the community, not making a living.

 

[HiTech]

By end of 2014, we are talking about at least 60,000 Teslas. Then it'll go up by at least 50-100k per year.
Also, for anyone that has an app already built, this would be some modification only. At about $5/app, you are talking about an opportunity of upto $250-500k/yr. Not a bad living at all!

 

[richkae]

By end of 2014, we are talking about at least 60,000 Teslas. Then it'll go up by at least 50-100k per year.
Also, for anyone that has an app already built, this would be some modification only. At about $5/app, you are talking about an opportunity of upto $250-500k/yr. Not a bad living at all!

The top free android apps have 500 million + downloads.
The top paid android apps have 1 million + downloads.
The market shows that less than 0.2% of android users will pay for the top "must have" app for a couple of bucks.
You think that 100% of Tesla owners will buy any app? I think if you mange to sell an app to 2% of Tesla owners you would be doing incredibly well, but not well enough to make a living.

 

[dirkhh]

I'm a little tech-stupid so, break it down for me:

Are these things that could be updated and used on the EXISTING car or would this require a different head unit? (i.e. retrofit being either impossible or expensive)

This is purely a software update. The head unit already is running Linux (and Tesla is violating the GPL while they are at it - but they are working on fixing this). So this is "simply" adding a different runtime environment and sandboxing it appropriately.
So technically all this isn't that hard. The fact that mister "optimism personified" gives you a one year time frame for it, with so many much more critical things already queued up and taking way longer than promised, should make you skeptical as to when realistically to expect this. Or if at all.

 

[AnOutsider]

The top free android apps have 500 million + downloads.
The top paid android apps have 1 million + downloads.
The market shows that less than 0.2% of android users will pay for the top "must have" app for a couple of bucks.
You think that 100% of Tesla owners will buy any app? I think if you mange to sell an app to 2% of Tesla owners you would be doing incredibly well, but not well enough to make a living.

Plus the base gen 3 vehicles may not have the screen and apps

 

[Gizmotoy]

This is purely a software update. The head unit already is running Linux (and Tesla is violating the GPL while they are at it - but they are working on fixing this). So this is "simply" adding a different runtime environment and sandboxing it appropriately.
So technically all this isn't that hard. The fact that mister "optimism personified" gives you a one year time frame for it, with so many much more critical things already queued up and taking way longer than promised, should make you skeptical as to when realistically to expect this. Or if at all.

We actually don't know that yet. Elon's quote said they needed to make some upgrades before they could move forward, but didn't go into details. It could be all software, as you mention, or it could require more powerful hardware. We just don't know.

Plus the base gen 3 vehicles may not have the screen and apps

That seems pretty unlikely, given most current EVs have at least small touch screens. The Gen 3 might not have a 17" screen, but it'll have something. Hell, I've been in rented Ford Escapes and Chrysler 300s that have touchscreens with apps in the past month (not that they were any good, but still). And if they're smart they keep the same app SDK to expand their installed base and lure developers.

 

[AnOutsider]

That seems pretty unlikely, given most current EVs have at least small touch screens. The Gen 3 might not have a 17" screen, but it'll have something. Hell, I've been in rented Ford Escapes and Chrysler 300s that have touchscreens with apps in the past month (not that they were any good, but still). And if they're smart they keep the same app SDK to expand their installed base and lure developers.

Maybe so. Elon did say maybe one screen instead of 2. In which case, they'd be going with physical dials on the screen behind the steering wheel to make the center screen happen.

 

[dirkhh]

We actually don't know that yet. Elon's quote said they needed to make some upgrades before they could move forward, but didn't go into details. It could be all software, as you mention, or it could require more powerful hardware. We just don't know.

Correct. We don't know. But some of us here know the software stack fairly well, have a solid understanding of the hardware that's in the main computer, and have experience how something like this would be architected (the wind speed information comes from a server - so this only works when you are connected, so why not do the compute intensive part of this on a server, anyway?).
What I'm saying is "it is quite straight forward to do this, based on the information that we have, with the existing hardware and just a software update". Literally - all I need is an API that gives me the destination the user entered and I could pretty much do this server side today. Give me a second API that allows me to feed a route into the navigation system and we are done.

This is NOT rocket science.

 

[Stoneymonster]

Literally - all I need is an API that gives me the destination the user entered and I could pretty much do this server side today. Give me a second API that allows me to feed a route into the navigation system and we are done.

This is NOT rocket science.

I'd go further and say that doing any of this stuff locally would be a colossal mistake.

 

[evme]

My point is that I don't want ported android apps running in my car. In general, I find android apps to be poorly-written and unstable to begin with (not to mention unsecured). Sandbox or not, that just doesn't seem like a great thing to have on the 17" dash -- especially when Tesla has put a lot of effort into the usability and UI of the current apps. Have you experienced ported android apps on BB10?

Just like the current apps, I'd prefer it if Tesla either paid companies to build specific apps for their platform or built a new wave of apps in-house. Sure, that may mean fewer apps, but I'll talk quality over quantity any day.

It is not like they are forcing you to use them. Using BB10 is not a good example because BB10 uses QNX. Tesla uses Linux in their infotainment system, this makes mapping of functions 1 for 1. It would be no different than running Android on an Android device. Android apps as of 4.X should be coded to work on any screen size. As far as stability and coding goes, the only apps that have those issues tend to be ports from other platforms. A native Android app should work just fine.

Would it be awesome if they make native apps for Tesla? hell yeah. But I'd rather have some apps in some way then no apps at all.

On top of that, it seems other car companies also plan to use Android such as Kia and Hyundai with their UVO 2.0. If Android does take off in more and more cars, app developers will most likely be more car aware.

 

[Gizmotoy]

Correct. We don't know. But some of us here know the software stack fairly well, have a solid understanding of the hardware that's in the main computer, and have experience how something like this would be architected (the wind speed information comes from a server - so this only works when you are connected, so why not do the compute intensive part of this on a server, anyway?).
What I'm saying is "it is quite straight forward to do this, based on the information that we have, with the existing hardware and just a software update". Literally - all I need is an API that gives me the destination the user entered and I could pretty much do this server side today. Give me a second API that allows me to feed a route into the navigation system and we are done.

This is NOT rocket science.

Despite being terribly condescending, I'll reply. Elon said he wanted to emulate and sandbox Android, and that apps would run there. We know that the existing Tegra3 barely has enough power to push the main display given the widely-discussed shortcomings of its 32-bit memory bus, and can even be demonstrated somewhat by the occasional choppiness of the new direction-up map orientation that requires lots of transforms, or scrolling in the browser. I wouldn't be surprised if a more-powerful infotainment system was in the works to support this effort, which may be backed up by Elon's comments that upgrades will be required.

The example you give is trivial, and something that's already stated to be included in a coming update. Of course that's easy.

Maybe we're not on the same page because you're talking about how you'd like it to work, and I was just going by how Elon said it would work? I don't know. I certainly agree a native API would be best, and having the option of both native apps and emulated Android apps would be the best of both worlds for Model S owners until a base of apps is built up.

I'd go further and say that doing any of this stuff locally would be a colossal mistake.

And yet that seems to be exactly what was proposed. There's no point in emulating Android's stack if the goal wasn't to run pre-existing apps locally. Otherwise, they'd just develop a fully-native API and let developers work directly with sandboxed Linux/Qt, completely bypassing Android and all the overhead its subsystems will introduce.

 

[AnOutsider]

It is not like they are forcing you to use them.

I'm aware. However, if they push these apps instead of purpose-built apps, that would likely mean that there would be less of the quality purpose-built apps that those who don't want android rehashes prefer.

Using BB10 is not a good example because BB10 uses QNX. Tesla uses Linux in their infotainment system, this makes mapping of functions 1 for 1. It would be no different than running Android on an Android device. Android apps as of 4.X should be coded to work on any screen size. As far as stability and coding goes, the only apps that have those issues tend to be ports from other platforms. A native Android app should work just fine.

The reports were "emulation". BB10 runs android apps in a sort of emulated mode. Tesla wants to emulate an android environment for the apps to run in (it's more than just being linux). The comparison is 100% spot on IMO.

Would it be awesome if they make native apps for Tesla? hell yeah. But I'd rather have some apps in some way then no apps at all.

That's your preference, and that's fine. As you'll note, I stated that _I_ didn't want them.

On top of that, it seems other car companies also plan to use Android such as Kia and Hyundai with their UVO 2.0. If Android does take off in more and more cars, app developers will most likely be more car aware.

Purpose-built apps for cars (whether emulated in an android environment or not) would be much more preferable to me than what is being talked about here (existing phone and tablet apps being run on the screen in an emulator).

 

[Stoneymonster]

And yet that seems to be exactly what was proposed. There's no point in emulating Android's stack if the goal wasn't to run pre-existing apps locally. Otherwise, they'd just develop a fully-native API and let developers work directly with sandboxed Linux/Qt, completely bypassing Android and all the overhead its subsystems will introduce.

Sorry I thought we were talking about a specific example of doing sophisticated route and consumption estimation on the local CPU.

 

[Gizmotoy]

Sorry I thought we were talking about a specific example of doing sophisticated route and consumption estimation on the local CPU.

In that case, I misread your statement. Sorry about that. For that particular case, doing things server-side can make a lot of sense. On the other hand, adding wind direction and elevation to route computation is not terribly more complicated that route computation by itself (distance/speed factors on a given potential leg are now distance/speed/wind/elevation). It seems to be the case that both local *and* server route computation is happening currently (presumably why the Navigon-based navigation system and Google Maps on the main display sometimes provide different directions), so it's probable current local route computation is feasible.

 

[evme]

I'm aware. However, if they push these apps instead of purpose-built apps, that would likely mean that there would be less of the quality purpose-built apps that those who don't want android rehashes prefer.

I don't Tesla will push them, they are offering it as an option.

The reports were "emulation". BB10 runs android apps in a sort of emulated mode. Tesla wants to emulate an android environment for the apps to run in (it's more than just being linux). The comparison is 100% spot on IMO.

Having 20 years of programming experience, I am going to have to disagree. Here is the thing, when you emulate code on different platforms. the code varies by platform as each platform behaves differently. Some functions may exist on 1 platform and not on the other and even the same functions can behave completely different. To explain it in a simple way, think of it like translating english to japanese, even if you translate it fully, it will not always have exactly the same meaning. A good example is word puns which effectively can't be translated. This is the difference between emulating android in linux where it is like translating english to english. In comparison to BB10 which is like translating english to japanese.

Purpose-built apps for cars (whether emulated in an android environment or not) would be much more preferable to me than what is being talked about here (existing phone and tablet apps being run on the screen in an emulator).

Well so far it seems Tesla, Kia and Hyundai are in. Doing a quick Google also points Renault uses Android for R-Link and Saab does as well. This hopefully will lead to more purpose built apps.

 

[vfx]

Tesla can vet the Tesla car apps like Apple does. Probably go all the way to Elon. No crappy apps for his car.


And for those that want to watch DVDs you can always plug a player into the Rear camera port screen.

 

[AmpedRealtor]

I hope that if Tesla implements an app store for the Model S, the apps are curated and held to a very high standard by Tesla. Most of the apps in the Android marketplace are crapware. Apple's app store model is far more successful and much more engaging. Seeing as how there probably would be more synergy between Tesla and Google than there would be between Tesla and Apple, I would expect more Google technologies to be implemented.