Automotive is one of the many field heavily dependent on software and unable to create anything good in that. That's a good example of why we damn need to change the IT landscape for a future society...
FLOSS MUST BE BY LAW, we can't accept a modern society of black boxes and this allow for collaborative development witch anyone own it's own while all other can cherry pick creating a fast enough, stable enough, safe enough stream of innovation.
Another part is electronic witch we are still champion but many industries who need electronics are TERRIBLE at it. Again open hw is mandatory to create evolvable architecture for any field.
We can't accept things like a car briked because parked in at a certain steepness and for some reasons a crapware update fails under that condition, overnight, without any human owner control. Similarly we can't accept safety features like ADAS being tied to car infotainment just because the automaker bought some ancient SoC and tablets for that purpose not knowing what to do otherwise. FLOSS prove to be a technical success, a business failure. It's time to make it the norm so business can compete in hw not in selling services.
Of course it's terribly hard but if back then at Xerox this happen it can happen again, recreating some relevant conditions like public fund for public FLOSS projects, created to serve the people not some private parties and used by any private party for commerce. This allow for a balance between the private sector needs and the public allow freedom of innovation and reducing certain level of crap for all.
If you have worked on automotive projects you know they follow some kind of standards like DLT for diagnostics and logging or at least they used to. A lot of it has moved to Android Automotive now.
Free software can be terrible as well, it solves nothing.
What you want to achieve with safety critical systems is diametrically opposed to common open source development patterns. A car company can not possibly take liability for a car where some random guy has flashed some software on.
> Free software can be terrible as well, it solves nothing.
Yes, it can, but being open, with an open toolchain, public from the start where the repo is just few SLoC, chance exists that someone else fix or even substitute it, so if a car sell because the hw/the design is good but software is bad chance exists that others fix it or even change it. Since ANY OEM of course try to grab as much as possible bad things likely disappear quickly while good one flourish. We might have terrible Android crappy ROM but also others totally different and being open it's hard to make them non interchangeable, meaning that after an initial phase there will be no VW or Rivian sw but common "distros" compatible with various cars, certified by the public to be safe for public use on the road.
Developing software for highly integrated embedded systems requires hundreds of thousands in validation, verification and testing hardware.
Wether the hardware and software are open is totally irrelevant. You can't have outsiders develop for it.
>it's hard to make them non interchangeable, meaning that after an initial phase there will be no VW or Rivian sw but common "distros" compatible with various cars, certified by the public to be safe for public use on the road.
This is completely delusional. This is not how hardware or software works at all.
Unless vehicles start to be kind of standard, like most Chinese EVs... Or like Japanese house abacus to speed up anti-seismic computations.
In the end we already have some common base, at least in EU most light commercial vans for instance have a common base even if they are built by many OEMs and have various cosmetic differences. A Transit, Vito, Ducato, Master etc are the same base with different dress, I see no reasons not to do the same for cars.
>Unless vehicles start to be kind of standard, like most Chinese EVs...
That is complete nonsense. In the Chinese market companies differentiate themselves by software.
I think you don't know what software in cars even is, so you are completely missing the point.
These common bases have been proposed, BMW at some point wanted German OEMs to cooperate on that, but they will never operate as open source projects, because they require hundreds of thousands in hardware to develop for. It is a massive project, which requires significant capital, OEMs can cooperate (which has a 99% chance of ending in total disaster), but every incentive they have goes against sharing.
About Chinese EVs yes, they show different software, but that's at the infotainment level, car core systems are essentially interchangeable and so are many parts, you can find the same on a SAIC, BYD, CATL, BAIC, Xpeng, Geely, Voyah, Chery, Changan and so on.
> I think you don't know what software in cars even is
Not in details, but I see the same pattern, a "core" part, with some buses and standards, mostly canbus and ODB, to communicate between core elements and in infotainment part in a tablet, hooked to the core for simple communications. Since all functions are pretty the same for every car because you can let's say design gazillion of breaking techniques, differentiating the break force/pulse per wheel, with various sensors to choose how and so on, the devices you actuate and the sensors you use are essentially the same the same. Of course at a very low level you are tied to the specific iron, like inside a CPU, but just above there is so little difference that you can have a common software stack. Failing to have one is a way to keep costs high while China get ground even if they are terrible at software.
>Then it's the time for a public funded and steered project.
Which means there is a 0% chance of a usable product. No manufacturer will ever use it and it will be all around terrible.
If VW is bad at software, the German government is at a totally new level. Btw. I have seen entry level software developer jobs for the German government starting at 35k.
>car core systems are essentially interchangeable and so are many parts, you can find the same on a SAIC, BYD, CATL, BAIC, Xpeng, Geely, Voyah, Chery, Changan and so on.
Wtf are you talking about. No, that is not true at all.
>Since all functions are pretty the same for every car because you can let's say design gazillion of breaking techniques, differentiating the break force/pulse per wheel, with various sensors to choose how and so on, the devices you actuate and the sensors you use are essentially the same the same.
You know that cars are different, right? Different systems, different behaviors, different sensors.
> just above there is so little difference that you can have a common software stack.
What are you on about?? No that isn't true at all.
>Failing to have one is a way to keep costs high while China get ground even if they are terrible at software.
How can the same software run on two cars which have totally different behaviors and components. Do you think you can run the same software on a rear wheel drive ICE with active suspension as on a all wheel EV with torque vectoring? This is delusional, you don't understand cars at all.
> Which means there is a 0% chance of a usable product. No manufacturer will ever use it and it will be all around terrible.
Xerox PARC products actually have had no formal commercial success, but the entire modern IT was born on them. To be more precise 80-90% of anything we have in any sector is born out of public research... Yes, the public does not produce nor sell products, that's a private sector thing, the public do research.
> If VW is bad at software, the German government is at a totally new level.
The public is not the government. German government have much more resources than VW and have all German public universities and research centers, who actually are much more skillful than VW, in essentially any fields.
> Wtf are you talking about. No, that is not true at all.
That's what I see from various Chinese EVs, well, three manufacturer at least, but that's is.
> You know that cars are different, right? [...] Do you think you can run the same software on a rear wheel drive ICE with active suspension as on a all wheel EV with torque vectoring?
Yes. As we have GNU/Linux on embedded systems, servers and desktop, the codebase is one, the compiled/configured final binary of course differ. Consider a thing: today it's cheaper putting an ESP32 in a roller shutter remote control than create from scratch a far simpler one. As I said specific devices have to be tied to the iron, but they can be managed with a common system and many functionalities can be put in the common system.
Something actually happening with ADAS typically handled by the infotainment system simply because they need much more computational horsepower than classic ABS, SRS and so on.
Stellantis actually try to do that concentrating in a SoC running QNX most of the car functionalities.
This is so tiresome. I thought you were talking about FOSS, but now you are telling me car manufacturers should do exactly what they are doing now, except the idiotic government should do it, so they can fail like Xerox Parc at commercial success. What?
>Xerox PARC products actually have had no formal commercial success
Which means that the example was a total failure, since the goal would be a commercial success, not a proof of concept leading to drastic decline.
>German government have much more resources than VW
The German government has definitely far less competent engineers. Which seem to be a somewhat valuable resource in building cars.
>German government have much more resources than VW and have all German public universities and research centers, who actually are much more skillful than VW, in essentially any fields.
Germany's most capable research organizations are mostly private. And are financing themselves by doing research for the German industry. I have no idea why you mention this though, treating software platform development as a research project is absurd.
>who actually are much more skillful than VW, in essentially any fields.
In any field except designing and producing cars at large scale.
>As we have GNU/Linux on embedded systems, servers and desktop, the codebase is one, the compiled/configured final binary of course differ.
You don't run Linux on an ABS system. That's, pardon the word, retarded.
>As I said specific devices have to be tied to the iron, but they can be managed with a common system and many functionalities can be put in the common system.
Sure, that is an E/E platform, which every car maker already has. Why another one?
>Something actually happening with ADAS typically handled by the infotainment system
Legitimately don't know what you are talking about. You know that cars have multiple processors for that exact reason?
>Stellantis actually try to do that concentrating in a SoC running QNX most of the car functionalities.
VW has the exact same concept. As does every car maker.
Let me hazard a guess - not specific to Volkswagen:
Awful, slow, buggy, unresponsive, unergonomical, poorly-integrated, occasionally dangerous software, including the user interface, produced by a company culture that just won't change as it just doesn't see value in this, therefore either outsources it to the lowest bidder or mismanage it internally.
There's a reason Android Auto and Carplay are selling points.
>produced by a company culture that just won't change as it just doesn't see value in this
I don't think that is the real problem. The issue is that Volkswagen has always been a hardware company and you can not make hardware like you do software. Fast iterations are not possible in hardware, so you need a careful long iterative process to fix previous problems and produce improved designs.
Implementing a software process into this can only lead to disaster.
They also know that this is a major problem, which is why they founded Cariad to develop software, but that got burdened by the same problems.
Also: compared to the US, software engineering is payed terrible in the EU. If you are writing software you don't get payed more than an engineer.
During the ICE era VW bought head units from Blaupunkt and Fuel Injection systems from Bosch and A/C systems from Delphi no need for a real software culture just integration expertise.
With EV's it is different ball game where software is a differentiator thanks to Telsa and the type of customers (techies) who are early adopters.
Anything developed using AUTOSAR is going to be awful shit. No competent programmer will willingly get within 10 feet of an AUTOSAR job posting if they've ever had to suffer through using it, and all the old ICE manufacturers standardized on AUTOSAR. They can't retain good programmers, and they can't hire programmers who have experienced their shit before, so they're left with naive newbies or overpriced contractors whose only goal is to milk the company for as many billable hours as possible.
FLOSS MUST BE BY LAW, we can't accept a modern society of black boxes and this allow for collaborative development witch anyone own it's own while all other can cherry pick creating a fast enough, stable enough, safe enough stream of innovation.
Another part is electronic witch we are still champion but many industries who need electronics are TERRIBLE at it. Again open hw is mandatory to create evolvable architecture for any field.
We can't accept things like a car briked because parked in at a certain steepness and for some reasons a crapware update fails under that condition, overnight, without any human owner control. Similarly we can't accept safety features like ADAS being tied to car infotainment just because the automaker bought some ancient SoC and tablets for that purpose not knowing what to do otherwise. FLOSS prove to be a technical success, a business failure. It's time to make it the norm so business can compete in hw not in selling services.
Of course it's terribly hard but if back then at Xerox this happen it can happen again, recreating some relevant conditions like public fund for public FLOSS projects, created to serve the people not some private parties and used by any private party for commerce. This allow for a balance between the private sector needs and the public allow freedom of innovation and reducing certain level of crap for all.