Honestly it makes no sense to show it as a full-dash replacement. That means the car has to ship with an iOS device or the owner is required to always have one plugged in?
I see this more as "here's some early Apple Car UI that you all can play with and tell us what works and what doesn't, and maybe one of you developers in the audience will make something interesting we can eventually steal".
The concept images are compelling but you won't see this in any new car any time soon.
I'm an automaker and I want a fancy UI on my dash (like Tesla or EQS). So I put in the big screen and then add the Apple extensions to allow CarPlay 2.
But what do I ship with the car? A cheap and quick UI with basic features, or do I have to write a second fancy UI that looks good in the showroom, on test drives, and when the battery dies in the phone? How much time and money do I spend on that?
If I'm Toyota and shipping cheap Corollas, what kind of screen do I pay for on that car? Triple-wide LCDs aren't exactly a commodity part. So is CarPlay 2 something only for expensive cars?
I wonder if Apple can strong-arm themselves into doing so. "Dear car manufacturer, [insert high percentage here]% of people who buy your cars use iPhones. If you want iPhone connectivity, we want the deep integration, or you and your customers get nothing, and they'll buy [competitor brand] instead.".
On the one hand, Apple are significantly better at design than car manufacturers so this should be good. On the other hand, moving from something at least close to an RTOS to this does concern me. When your speedometer freezes, who is going to take the blame?
Yep I'm curious what mode it goes into when the iPhone disconnects. Presumably there is some built in local processing in the car's "display" which continues to render the important bits of the UI without connected features and widgets?
Firstly, there's a very good chance driving without a speedometer is illegal in your area.
Secondly, anything that affects functional safety should itself be vetted fron a FuSa perspective. I guarantee you that most manufacturers treat the instrument cluster as a Functional Safety item. If CarPlay can't bring the system to a state where that analysis is affected (which seems unlikely here), there's no issue.
You shouldn't crash, but it's not unheard of for the speedometer to fail.
Certain early-2000s GM speedometers were prone to breaking. Either the servo motors straight-up stopped working, or the label started peeling up and jammed the needle. My brother had several trucks from that era with this problem.
I don’t know what they could be doing differently when CarPlay is activated, but I’ve never had it crash on me, and I’ve been using it almost daily for 3+ years.
The iPhone on its own has locked up a handful of times, as has the built-in VW entertainment system.
If your speedometer freezes on Carplay, I assume you can just unplug iPhone and get stock car speedometer. I would be concerned if iPhone controlled breaks, gas pedal or wheel but informational displays are fine.
In addition, I have VW Tiguan with glass cockpit (or whatever you call it when it does not have mechanical speedometer) for about 3 years and it froze on me a couple of times. The only way to fix was to turn off/turn on the car.
Yeah, I'm so not into this. While I would love a more visually compelling and useful in-car infotainment system, something being tied to iOS or Android is a hard no from me. I don't trust Apple to be anything other than a closed ecosystem, and I don't trust anything by Apple or Google to made obsolete well before the end of a vehicles useful lifetime.
And you trust the car manufactures to write good software? The one reason I wouldn’t buy Tesla is because of its infotainment system.
The on device display is basically a dumb remote terminal. Those are easy to stay compatible forever. I can still AirPlay from my an old 1st gen iPad to a new Roku TV. I can still print from an old IPad to a new printer.
Car manufacturers often build software using an entirely different set of constraints: Real-time operating systems with extremely reliable code. Because when they fail, the vehicle recall often involves hauling every car into a dealership and having them flashed. The occasions these fail lead to scenarios like "the car accelerates suddenly" or "the car abruptly stops in the middle of the road".
The fact that Apple and Tesla aren't good stewards of mission critical car software is, in fact, the fact it's trivial to push updates. It means they can shove out stuff with less validation and fix it in post. I would contend that easy patching nearly guarantees poorer software development correctness.
Consider the fact that no Nintendo GameCube game has to be post-launch patched, and while some flaws do exist that speedrunners can exploit, all of the games still work, decades later. Compare that to modern games, which often require a "day one patch" to fix them.
I absolutely trust a traditional car manufacturer to release better software than Apple or Google. It may not be as pretty or have as many bells and whistles, but it probably isn't going to kill me.
> Compare that to modern games, which often require a "day one patch" to fix them.
It could be argued that this is for slightly different reasons - because of the delay between a game being submitted to console manufacturers for testing and release, and the "crunch time"-style industry of game dev, I'd argue this is mostly to try and get the game launched sooner, because its cheaper/faster to ship a buggy game on disc and fix it on release day than to wait until the game is entirely fixed first, now that it is a viable option
(Not to disagree with your point - I completely agree)
This is half accurate. CarPlay doesn't control anything directly, true. However, CarPlay is providing information that users need to drive safely, and that can probably fail spectacularly. So your claim that the car isn't going to crash because Apple Music is flaky is false.
I wonder if this will be the default interface for the car (with iOS devices the only first-class citizens, and very limited Google offerings) or if its going to be a mode that gets activated when an iPhone is present...
> I don't trust anything by Apple or Google to made obsolete well before the end of a vehicles useful lifetime.
Car dashes should just have an DisplayPort or USB port we plug chromecast-like or a hypothetical apple-cast-like device into, imo. And those can mediate the experience with our phones/devices, but are themselves affordable & replaceable.
This would be much more sensible if there were a coherent, sensible, reasonable, fair reasonable and non-discriminatory standard for IoT that was available, for communicating things like sensors, camera views, dealing with multi-zone audio, setting brightnesses/heat, adjusting chair shape... Right now we're still in goofy world where everything is ultra-proprietary and there's not even a common base we hypothetically could extend/build around.
> IoT that was available, for communicating things like sensors, camera views, dealing with multi-zone audio, setting brightnesses/heat, adjusting chair shape...
There are practical considerations. You need top end device with mobile connections to drive experience similar to Carplay or Android Auto with fancy map animation, fast touch response time and music streaming apps. That device cost at least $700+. Nobody wants to use slow as death chromecast like IoT to drive car experience when you have phone in your pocket.
Why build something when there is no demand? Besides there are all sorts of security consideration in not allowing random devices to change to many things in the car.
> You need top end device with mobile connections to drive experience similar to Carplay or Android Auto with fancy map animation, fast touch response time and music streaming apps. That device cost at least $700+.
I'm pretty sure a boring $50 Chromecast could drive this experience just fine & without challenge. An 8-core ARM chip with a decent GPU is remarkably affordable in this day & age, and more capable than most people realize.
It would be significantly different to try to drive multiple experiences at once, versus a Chromecast which runs a single app at a time. But there are just so many cores now. Even if they're low performance Cortex-A510, I don't see the need for big CPU. And nothing I've seen on any of these videos gives me the feeling big complicated GPUs are necessary: the whole thing experience is powered by pretty lo-fi simple Widgets, "glancable" experiences.
8 core ARMs aren't typically automotive grade and that's the expensive bit. An automotive grade iMX.8 with 4 A53 cores runs somewhere around $100. You could step down to the old iMX.6Qs (the product line your infotainment probably already uses), but they're still $60ish/ea.
> 8 core ARMs aren't typically automotive grade and that's the expensive bit. An automotive grade iMX.8 with 4 A53 cores runs somewhere around $100.
I think we're into splitting hairs territory. You're right, automotive industry is typically unbelievably crazy far behind, and accounting for that is a pain.
ARM in general has made me very very very sad for a decade of non-delivery now. Cortex A53 turns 10 years old in October. ARM still doesn't have any replacements actually shipping that are really any better- A55 was like a 5-15% boost (a real stinker), A510 isn't actually shipping yet, & who knows if it'll really be better this time, another sad repeat of A55.
But I don't think it changes my arguments or belief significantly. We could easily not-suck if we wanted to. Some big car company could go work with a foundry or hyperscaler & decide to kick out a lower-middle speed automotive ARM chip, which would cost very very little. I think it's also conceivable that simply using an extended temperature range chip & being a lot more careful about power regulation & noise tolerance is perfectly doable.
Let's look at the parent:
> You need top end device with mobile connections to drive experience similar to Carplay or Android Auto with fancy map animation, fast touch response time and music streaming apps. That device cost at least $700+.
If we evaluate your critique a little more, look around, we can go further with your arguments: anything decent and Automotive-grade is not available, at any price. No one cares, no one buys it, no one makes it. Right now.
Anyhow I was just throwing out numbers. You're right, we might not have 8 cores to throw around right now. Maybe only 4. A Google TV runs off a Allwinner S905D3 with slow quad A55 and low end G31MP2. It still runs a decent 4K, is pretty fast & responsive (hindered largely by flash speed). It's barely warm to the touch. The gist was: even basic cheap computers are quite fast & capable, and you're pointing out that automotive prices are high. I have to add that there are no high end cores for automotive right now, at any price, and generally just automotive is vastly under-served. But I think, ultimately, there's probably still plenty enough oomph to run a system like this, with whatever we've got.
I'm not as optimistic as you about the technical capabilities of OEMs and Tier 1s at designing good silicon and software. They're monstrously good at producing millions of thingsthat will survive the most horrendous conditions imaginable at surprisingly economical prices, but terrible at modern electronics and software. That inability keeps me employed, so slight bias here.
As for simply using extended range components, Tesla actually did that for several years of the Model S infotainment system. Turned out that wasn't enough and the components started failing after a couple years [1].
The screen is 100% not what I would skimp on, as they have immense mechanical wear applied to them in a car. Tesla also famously uses AMD processors in their cars; unless it's a well kept secret, AMD doesn't make any automotive parts. I would not be surprised to see ruggedized electrical defense combined with consumer application processors be fairly widespread in today's vehicles.
Apple's proposing putting widgets in the center console. There's notably already a variety of widget games[1]. Apple does spend over a minute of their WidgetKit introduction in WWDC2020 declaring "widgets are not mini apps" but already this is a blurred line. I'd be more excited to see how Apple handles this situation, except I rather expect Apple just to be Apple and to manually approve each car widget.
Nope, absolutely hard no for me. As long as I can play my own music and answer calls in my car, I’m happy. I don’t need any of this shit on my dash as I’m not already spending way too much time looking at my phone screen anyways. Why would I want my passengers to see what’s on my calendar or see my messages? Its actually ridiculous.
There has to be a backup display that is not iOS (Apple - whatever). Cars last 10 or more years. Will Apple support this for at least that long? Also, in 10 years is any who doesn't own an Apple device going to be able to buy and drive a used car with this?
My vehicle has a HUD that includes navigation hints when using the built-in GPS. It doesn't work when using Android Auto or Apple Carplay. It would be nice to have deeper integration like this.
44 comments
[ 2.7 ms ] story [ 60.6 ms ] threadThey're probably working on their new monthly subscription plans for "activating" this feature...
I see this more as "here's some early Apple Car UI that you all can play with and tell us what works and what doesn't, and maybe one of you developers in the audience will make something interesting we can eventually steal".
The concept images are compelling but you won't see this in any new car any time soon.
Alternatively a car could just come with the guts of an Apple TV built in to run an unconnected version of CarPlay when an iPhone is not connected.
I'm an automaker and I want a fancy UI on my dash (like Tesla or EQS). So I put in the big screen and then add the Apple extensions to allow CarPlay 2.
But what do I ship with the car? A cheap and quick UI with basic features, or do I have to write a second fancy UI that looks good in the showroom, on test drives, and when the battery dies in the phone? How much time and money do I spend on that?
If I'm Toyota and shipping cheap Corollas, what kind of screen do I pay for on that car? Triple-wide LCDs aren't exactly a commodity part. So is CarPlay 2 something only for expensive cars?
Secondly, anything that affects functional safety should itself be vetted fron a FuSa perspective. I guarantee you that most manufacturers treat the instrument cluster as a Functional Safety item. If CarPlay can't bring the system to a state where that analysis is affected (which seems unlikely here), there's no issue.
That said, I'd probably use it.
Certain early-2000s GM speedometers were prone to breaking. Either the servo motors straight-up stopped working, or the label started peeling up and jammed the needle. My brother had several trucks from that era with this problem.
The iPhone on its own has locked up a handful of times, as has the built-in VW entertainment system.
In addition, I have VW Tiguan with glass cockpit (or whatever you call it when it does not have mechanical speedometer) for about 3 years and it froze on me a couple of times. The only way to fix was to turn off/turn on the car.
The on device display is basically a dumb remote terminal. Those are easy to stay compatible forever. I can still AirPlay from my an old 1st gen iPad to a new Roku TV. I can still print from an old IPad to a new printer.
The fact that Apple and Tesla aren't good stewards of mission critical car software is, in fact, the fact it's trivial to push updates. It means they can shove out stuff with less validation and fix it in post. I would contend that easy patching nearly guarantees poorer software development correctness.
Consider the fact that no Nintendo GameCube game has to be post-launch patched, and while some flaws do exist that speedrunners can exploit, all of the games still work, decades later. Compare that to modern games, which often require a "day one patch" to fix them.
I absolutely trust a traditional car manufacturer to release better software than Apple or Google. It may not be as pretty or have as many bells and whistles, but it probably isn't going to kill me.
It could be argued that this is for slightly different reasons - because of the delay between a game being submitted to console manufacturers for testing and release, and the "crunch time"-style industry of game dev, I'd argue this is mostly to try and get the game launched sooner, because its cheaper/faster to ship a buggy game on disc and fix it on release day than to wait until the game is entirely fixed first, now that it is a viable option
(Not to disagree with your point - I completely agree)
Your car isn’t going to crash because Apple Music is flaky.
Car dashes should just have an DisplayPort or USB port we plug chromecast-like or a hypothetical apple-cast-like device into, imo. And those can mediate the experience with our phones/devices, but are themselves affordable & replaceable.
This would be much more sensible if there were a coherent, sensible, reasonable, fair reasonable and non-discriminatory standard for IoT that was available, for communicating things like sensors, camera views, dealing with multi-zone audio, setting brightnesses/heat, adjusting chair shape... Right now we're still in goofy world where everything is ultra-proprietary and there's not even a common base we hypothetically could extend/build around.
It gives me headaches just thinking of it.
Why build something when there is no demand? Besides there are all sorts of security consideration in not allowing random devices to change to many things in the car.
I'm pretty sure a boring $50 Chromecast could drive this experience just fine & without challenge. An 8-core ARM chip with a decent GPU is remarkably affordable in this day & age, and more capable than most people realize.
It would be significantly different to try to drive multiple experiences at once, versus a Chromecast which runs a single app at a time. But there are just so many cores now. Even if they're low performance Cortex-A510, I don't see the need for big CPU. And nothing I've seen on any of these videos gives me the feeling big complicated GPUs are necessary: the whole thing experience is powered by pretty lo-fi simple Widgets, "glancable" experiences.
I think we're into splitting hairs territory. You're right, automotive industry is typically unbelievably crazy far behind, and accounting for that is a pain.
ARM in general has made me very very very sad for a decade of non-delivery now. Cortex A53 turns 10 years old in October. ARM still doesn't have any replacements actually shipping that are really any better- A55 was like a 5-15% boost (a real stinker), A510 isn't actually shipping yet, & who knows if it'll really be better this time, another sad repeat of A55.
But I don't think it changes my arguments or belief significantly. We could easily not-suck if we wanted to. Some big car company could go work with a foundry or hyperscaler & decide to kick out a lower-middle speed automotive ARM chip, which would cost very very little. I think it's also conceivable that simply using an extended temperature range chip & being a lot more careful about power regulation & noise tolerance is perfectly doable.
Let's look at the parent:
> You need top end device with mobile connections to drive experience similar to Carplay or Android Auto with fancy map animation, fast touch response time and music streaming apps. That device cost at least $700+.
If we evaluate your critique a little more, look around, we can go further with your arguments: anything decent and Automotive-grade is not available, at any price. No one cares, no one buys it, no one makes it. Right now.
Anyhow I was just throwing out numbers. You're right, we might not have 8 cores to throw around right now. Maybe only 4. A Google TV runs off a Allwinner S905D3 with slow quad A55 and low end G31MP2. It still runs a decent 4K, is pretty fast & responsive (hindered largely by flash speed). It's barely warm to the touch. The gist was: even basic cheap computers are quite fast & capable, and you're pointing out that automotive prices are high. I have to add that there are no high end cores for automotive right now, at any price, and generally just automotive is vastly under-served. But I think, ultimately, there's probably still plenty enough oomph to run a system like this, with whatever we've got.
As for simply using extended range components, Tesla actually did that for several years of the Model S infotainment system. Turned out that wasn't enough and the components started failing after a couple years [1].
[1] https://www.thedrive.com/tech/27989/teslas-screen-saga-shows...
https://pbs.twimg.com/media/FUlm_XEUUAAViqE?format=jpg&name=...
[1] https://apps.apple.com/us/app/widget-games-casual-arcade/id1...
Let cars be cars for fucks sake.