If the two 2023 VW models I've driven are any indication, it's an abject failure.
It might be partially because I haven't seen what a "new" car looks like in a while, but I found everything horrible. Both the actual touch menu navigation as well as the applications - possibly the worlds most useless distance sensing / park assistance technology who's only mode of operation is to been insanely at you when you're near anything. I've also had it just crash and had to look up how to reset by holding the power button for a long time.
I think this stuff may be par for the course on new cars now, but VW in particular is shit.
I would add that I don't know what the design goals of the OS are. It appears that in-car OSs generally are for the utility of the car company, whether stealing personal information or upselling, nagging, or preventing you from doing things you'd like with your car. So maybe it's going great on those fronts.
Honestly, I have multiple friends who were Volkswagen fanatics for life, normal folk who would only consider a VW product for their next car, all of them said they will look elsewhere after test driving the latest generation of cars. The touch interface is actually insane. It's honestly designed by someone whose goal is "try to see what's the worst interface you can create and still sell a car". The capacitive window controls deserve some kind of award on their own. It's awful awful awful 1000x. A friend of mine is literally in the process of selling his nearly new Golf Mk8 over this shit - he said he thought he would get used to it, but it leads to misclicks and accidental activations of functions literally every day. He has the entire system crashing at least once a week
usually when driving. And multiple times the entire system just switched from miles to km on its own without any input. He reported it to VW who of course said they can't reproduce the issue. It's unusable. I wish reviewers would use extremely harsh words about it - it doesn't matter how well the car drives or how much space it has if the interfaces inside it are literally unusable.
I honestly recommend seeing how it's done as it begs to ask how this left the design stage without the head designer being fired and barred from the automotive industry forever. If the buttons to control all windows were all capacitive, that would be bad enough on its own, but perhaps manageable. But they decided to go with two buttons for left/right window, then a separate capacitive button that switches those other two between controlling front or rear windows. And no of course it offers no tactile feedback, so if you want to open the rear window you have to swipe on this button you can't even feel and that you can't see the state of without looking down, then you need to find the capacitive button for windows control to roll the window down. It's 100% insane. The only way to make it worse would have been to remove the buttons entirely and make it all voice controlled only.
Indeed, you’d think Mazdas take is so logical and natural in the context of car safety that all manufacturers would follow, but here we are, with new models moving away from any tactile control.
> I wish reviewers would use extremely harsh words about it
Some are. Drive Tribe on YouTube has been very critical of Ferrari’s move to use touchscreens entirely and while praising the cars themselves, they outright claim the UX is is simply unusable and they just couldn’t recommend them just because of it.
Porsche is part of the VW group, and while you could expect something better for the price paid, the UI is even worse in their EV Taycan. With poor touch screen feel and frequent crashes too.
Luckily they integrate, when it works, with CarPlay and Android Auto that provides a 2023 like interface.
The last year drama firing the CEO for many reasons, one being poor software results, doesn't bring hope for a fresh update any time soon.
It is a good car* imprisoned within an utterly abysmal user interface. I want to add one new data point to this conversation, the parking brake button, one of the four tactile buttons available, will actuate past detent and click on occasion without engaging the parking brake. This is incredibly dangerous and must be addressed (VW staff on HN please find a way).
Further, repeated lamentations. Any basic task, changing fan speed, turning on heated seats, looking at the battery charge level is dangerous in motion. The latency is absurd - mid-10s iPad on a 105 degree day under strain bad.
It’s also notable that the design of the touchscreen’s surrounds is poor. There is nowhere to place your hand so you’re groping around in the air without any tactile reference. It’s just really bad, I’ve never successfully activated the defogger without turning on something else by mistake.
I have another VW family station wagon where all of the controls are physical, including a nice overhead switch panel. It feels special and is easier to use in all regards. It is a charming and thoughtful piece of human-machine interface design.
I won’t abandon electric cars but I will pay close attention to physical button availability going forward.
*the brakes do not provide the authority I expect from a VW
Know I’m in the minority, but I think a separate parking brake button that arrests the car is a really important safety feature. In the other vehicle I can pull a switch and know that the car will not move.
What a strange puff piece, given how cynical and snarky the Ars house style is these days.
It's clear that this project is a mess, but most of the article is just some exec giving jargony but information-free quotes to make it sound like everything is going just great.
I feel like Ars carried the automotive tech torch from Jalopnik when they slid into obscurity after Gawker imploded.
The reporting has always been mediocre - there is so much interesting tech happening across the industry, yet they choose to report on puff pieces like this, or paid-for promos.
I wanted a VW EV, but the years-long laments by reviewers about how slow and unresponsive their screen software was kept me from buying one. I’ve owned a half-dozen new VW/Audi products over the years. Whatever they’re doing with software still, the brands have almost no appeal to me any longer.
Somehow Ford revamped Sync faster than VW has optimized their software.
With the 2022 update it's not as bad as some of the initial reviewers say, in the ID4
No problem with lag at all.
The nav and trip planner is actually way better than Google maps IMO. It will direct you where to charge and for how long on longer trips. The lane guidance with the light strip in the dash and directions in the gauge cluster are extremely useful.
It is annoying that media and climate controls are all on the touch screen. However I learned to just set it to auto and use the dedicated capacitive temp buttons to tweak a degree or two.
For radio and music, I almost always use the buttons on the steering wheel.
The driver assist features work very well on the highway, standard lane center and adaptive cruise.
The mobile app sucks and only controls charging and remote climate.
Overall a good practical car that's solidly put together. It's extremly quiet and has a shockingly good turn radius. My one complaint is that it's a bit bloated weight wise. VW is moving back to physical controls in their newest models, but North America seems to be destined to only receive big crossovers for now.
I have a very early ID.4 and I’ve yet to receive any update since taking possession. I did just get a safety recall notice, which as I understand it, once fixed will enable further updates.
If they could replace the infotainment software and re-design all of the knobs and buttons, I would give my ID.4 a 9/10. That said, I have a X4, too, and I prefer driving the ID.4.
I drive a 2023 ID.5 for 2 months and I have not faced many issues yet. Yes, the touch interface could be a bit more responsive, but I am overall very happy with the driving experience.
I believe Google is playing the long game in cars software. Car companies, and the incumbent car infotainment suppliers still believe they know what drivers want, when, in reality, almost all car infotainment is much worse than Android-based software.
There is too much capability and polished UI in smartphone-derived platforms. There are too many apps, especially media apps, that users want. Google maps and navigation is the "killer app" that no other car software maker can surpass.
Eventually, there will be two, maybe three holdouts who can justify vertical integration and everything else will be based on Android Automotive (not Auto!). In the slipstream of Android Automotive will be Waymo's autonomous driving software, plus the Waymo ride hailing UX. That, also, will become a fallback for everyone doing robotaxis who doesn't benchmark better than Waymo.
> I believe Google is playing the long game in cars software. Car companies, and the incumbent car infotainment suppliers still believe they know what drivers want, when, in reality, almost all car infotainment is much worse than Android-based software.
It reminds me of early cellphones. Functionally from an OS perspective there was little/no difference between a "flagship" device like the Motorola Razr and the free phone you could get at the time. Verizon essentially neutered any potential for differentiation and they thought they knew the experience and features users wanted. Apple basically came along and said no. It's pretty wild to me how bad car UI/UX is still especially given how expensive they are.
Cars are unlikely to go full open source, as there are safety requirements and so security by obscurity is a part of the plan (not the main part, but a last ditch prevention). ThThey will open source less critical parts.
Not open hardware, not user-upgradable nor -modifiable. There's a difference between Open Source and Free Software, and Android is a poster child for the former.
As someone not having dwelt in the automobile industry, I found this quote interesting:
> The idea is still right, but of course, as you have seen, we've gone through the storming phase, a norming phase, now you know, having to deliver in the performing phase, but it's what you would expect
I can guess what these phases are, but is that a common workflow in the industry? I never heard anyone to refer to these cycles in software development. For me it just sounded like a poor attempt to explain the delays with some buzzwords.
German car companies (or for what it's worth, everyone but Tesla) and their suppliers still operate in waterfall development. They never learned the lessons the software industry did, because (like other brands like Sony or Nintendo that are known for good hardware but bad software), they deem themselves hardware companies at the core as they haven't realized software is ruling the world.
Design something, ship it unchanged for three to four years, launch an incremental version, give another three to four years, ship a complete redesign. And that's not just valid for hardware but for software as well. Something like what Tesla does, changing stuff around whenever something is improving or they need to save money or whatever is completely antithetic to that.
Part of the reason is how Tesla operates: they manufacture everything they can in-house, the "dealerships" (=glorified showrooms) and service stations are all owned by Tesla. In contrast, the trad-car industry has soooo many layers with independent entities: the car manufacturers, their local sub-division fiefdoms, suppliers who are locked in with year-long contracts, brand-owned dealerships and service stations, independent dealerships and service stations... changes there take ages to be communicated across the field and become implemented. Tesla management, meanwhile, can just shoot off a weekly newsletter and everyone relevant now knows that starting from production date X there are only cameras but no ultrasound parking distance sensors, no matter the consequences.
> German car companies (or for what it's worth, everyone but Tesla) and their suppliers still operate in waterfall development.
That's always easy said but not true... sure all (VW, BMW and Mercedes) try to embrace agile (with the usual problems of a corp scaling it, and also a website is just different to the car system's) and also see software as what it is and wanting to pull more inhouse and try to unify platforms.. but also need to be realistic: those big ships move slow, you need to fight a lot "this has always been done that way" and those systems are just huge, etc ..
They may try to be "agile" on parts of their product development, I can confirm that from direct experience, but the fundamental product development culture is still waterfall.
> but also need to be realistic: those big ships move slow, you need to fight a lot "this has always been done that way" and those systems are just huge, etc
Indeed. The problem is, the competition is way faster. On the one side you have "established startups" like Tesla that have moderate-speed improvement culture, but on the other side you have China which is basically Tesla on steroids but with the utter financial might of the CCP behind it.
The old guard may go under simply because they thought for too long they could bribe Western politicians instead of improving.
What’s insane is that Tesla has completely wasted the software—based advantage it once had because Tesla management is so erratic. They’re wasting time on fluff and bizarre direction shifts from the top while VW and Hyundai and Ford and GM and Toyota are all either shifting to electric or already overtaking Tesla. Tesla’s only remaining moat is the charging network, and if that continues to be the case, the competition will eventually fix that too.
I was in the market for a new car in 2021… and tested a whole bunch of EVs.
All my previous cars were VW's. But the abysmal software was _the_ reason I didn't go with VW (or any of the other brands) this time.
Skoda Enyaq was the best of the bunch, the UI was fairly intuitive but still slow and unresponsive.
VW Id4 was over the line bad, but the worst of them all, it still annoys me to this day, was Audi Q4.
For a car with that price point, it was like the engineers made a sport of how bad they can make a UI.
A pop-up saying something about privacy every time you start the car.
Very regularly a pop-up saying "no data connection, while you could see LTE in the background, and everything appeared to just function normally.
And many, many more
The only usable part is the fact that you can use Android auto/carplay.
In the end I went with a Polestar, and honestly I am very pleased. Would recommend it to anyone.
Cars already have an operating system. It's a dash with dials you can look at to get info, a wheel you can use to steer, a set of pedals to control motion, a set of switches and buttons to control lights and such, and a center console with physical buttons that control extras like radio and heat/cooling. Not everything needs to be a computer!
I do some embedded systems work, albeit using it as a research tool and not directly involved in product development. Whether or not to have an OS is a perennial debate for projects that fall in between Little and Big. I turn to an OS when I expect to benefit from pre-written tools, such as file systems and higher level networking functions. I also expect portability of software from one hardware configuration to another.
For things that need to run in real time, or with super high reliability, but are relatively simplistic, goodbye OS. There's also a method where a computer with an OS is connected to multiple satellite computers that run simpler code. Turning a car into a network of tiny computers could reduce the sheer amount of wiring needed.
As mentioned elsewhere, using a mainstream OS taps into the skill set of people who know how and want to program against it. Also, the OS serving as a "layer" between applications and hardware lets you change hardware without having to rewrite all of your software.
There's a trend in every industry right now to turn everything into a computer. I can imagine the marketing meeting: "We need to get people to stop thinking about a car as a car. A car is software."
A modern car is a network of tiny computers, at least dozens, sometimes up to 100+. You can stretch that number significantly if you want to count every IC with an embedded microprocessor in it.
All of that diversity will be managed by a relatively small number of people at any OEM. Standardization to a minimal set of OSes and hardware platforms is the only way to make the workload manageable and keep the dozens of application teams unblocked.
For embedded C, RTOS by default is common. Not because applications need non-cooperative scheduling by default, but to manage memory-safe concurrency, and other abstractions. With rust, you can get away with simpler scheduling/locking mechanism and skip the RTOS until complexity reaches a threshold, or you have non-cooperative tasks.
Cars already are quite heavy computers and have been for ages. It's not like a new development. They have control units, component control units, the radio, bla bla bla. Lot's of integrated stuff.
Of course. But driving a car doesn't need a user-facing OS. It shouldn't feel like using a computer to drive a car. Am I misunderstanding the car operating system they were working on? I thought it was to facilitate UI on touchscreen controls not just low-level systems.
Cars have a number of processors, mostly running RTOSs, mostly controlling subsystems, like window motors, seats, etc. It's this way because these subsystems come from different suppliers. A lot of them do not need high stakes security hardening or hard real time capability. These will increasingly get integrated into software modules that can run in an evolved main "infotainment" system.
small car components are just MCUs, bigger units like ECU are large SoCs with full fledged RTOS like Integrity RTOS (which can be found on everything from cars to F-22 Raptor jet)
But industry standard is using an autosar based RTOS for all real time controller that you mention. Infotainment is mostly android or an OEMs private OS (which mostly means Linux with custom developed apps). Maybe some rare leftovers with QNX or WinCE.
A „headunit“ might consist of multiple CPUs, real-time and non real-time ones.
I doubt anyone would do a drivetrain controller on Linux or an infotainment system based on a RTOS.
Working with German engineers from Santa Clara time zone gives you only a few hours of overlap per day. That’s undoubtedly part of the problem here. There’s limits to how distributed a team can be.
Source: two years working from Bay Area with German and Swiss engineers.
It might be partially because I haven't seen what a "new" car looks like in a while, but I found everything horrible. Both the actual touch menu navigation as well as the applications - possibly the worlds most useless distance sensing / park assistance technology who's only mode of operation is to been insanely at you when you're near anything. I've also had it just crash and had to look up how to reset by holding the power button for a long time.
I think this stuff may be par for the course on new cars now, but VW in particular is shit.
I would add that I don't know what the design goals of the OS are. It appears that in-car OSs generally are for the utility of the car company, whether stealing personal information or upselling, nagging, or preventing you from doing things you'd like with your car. So maybe it's going great on those fronts.