Hacker News new | past | comments | ask | show | jobs | submit login
Mercedes-Benz previews its operating system MB.OS (mbusa.com)
186 points by mfiguiere on Feb 22, 2023 | hide | past | favorite | 386 comments



I was recently an owner of a Merced-Benz commercial van, and the software was by far the worst part of that whole experience.

It had obvious race-condition type bugs when it came to the user interface layer, but most frustrating was its tendency to succumb to some kind of memory leak on long drives where the entire head unit would just lock up and crash to a black screen after 6-7 hours of being turned on. Because the vehicle kept the computer system "warm" for up to 30 minutes or so to avoid doing a full (and slow) bootup process every time you stopped for fuel, this was a real problem on long trips and couldn't always be solved by power cycling the vehicle.

Had a dealer try to update it twice, which didn't seem to meaningfully impact the system stability at all.

Then there were multiple other, non-head-unit related glitches like the lane assist and cruise control features being incompatible with the state of Nevada (if the system fails to detect any other vehicles for a period of more than about 90 minutes, it assumes that there is a sensor fault, and refuses to operate [1]. Unfortunately it is quite easy to spend hours on the road alone in many southwestern US states, triggering this failsafe mode)

Suffice it to say that I am very skeptical of any software coming from Mercedes these days.

[1] https://www.winnieowners.com/forums/f265/2019-sprinter-cruis...


> if the system fails to detect any other vehicles for a period of more than about 90 minutes, it assumes that there is a sensor fault, and refuses to operate

I have a current model Sprinter and this behavior is frustrating (thanks for the explanation, by the way, I thought it was a random bug). Cruise will also disengage in heavy rain or snow, with the system saying the sensor is dirty. It wouldn’t be so bad if traditional cruise control would keep working without the active braking assist, etc. but unfortunately it’s all or nothing. I do like the system when it works, which is most of the time, and it’s never done something that felt unsafe.

I haven’t had any problems with the info system. This is my first vehicle with a touchscreen, and I was worried that it’d be a pain to use. But Mercedes provides real buttons for common actions, little trackpads on the steering wheel that can control the system without ever having to touch the screen, Car Play works great, and the built-in nav system is useful for those times I need directions and don’t have a cell signal.

My biggest worry is how these delicate parts will hold up over the years. My 10 year old car has a monochrome dot matrix display that loses a good number of pixels when it gets hot out. Will I really be able to get replacement parts for the Sprinter when the screen or computer start to die in 10-15 years?


> Cruise will also disengage in heavy rain or snow

That’s what it should do, honestly. It’s not safe to drive with cruise control in low traction conditions.


This is one unconscious bias from California-based self-driving companies: weather conditions worse than overcast skies exist in other parts of the world.

Recently, I drove in a whiteout blizzard with huge flakes without much accumulation on the road. Road was drivable, but I couldn’t see road markings. The experience would have been less scary with some sort of augmented display to show the road borders.


Unfortunately the augmented displays don't have better input and are likely much worse than your eyes (their major advantage is never getting bored or sleepy). In your described conditions it would likely have been even more blind than you were, so if there had been displays they would have little bearing to reality.


"This is one unconscious bias from California-based self-driving companies: weather conditions worse than overcast skies exist in other parts of the world."

Mount Baldy (right next to LA) is expected to see 7+ feet of snow over the next few days.

I'm fairly certain they're aware of the particular fact you mention. We do have (had thanks to this drought) multiple ski resorts across the state that do not rely upon artificially-made snow to operate.


https://telstarlogistics.typepad.com/telstarlogistics/2007/0...

> Developed by researchers at the University of California at Berkeley, the Advanced Rotary Plow (ARP) uses a network of magnets embedded along the roadway to guide the massive snow-spewing machine along the Interstate in zero-visibility conditions -- like a blinded pilot flying on instruments. Meanwhile, an onboard array of sensors sniffs out obstacles -- such as abandoned cars, hungry cows, napping yetis, or whatever -- that may lie hidden beneath the deep snows.

https://path.berkeley.edu/development-advanced-rotary-plow-a...

https://path.berkeley.edu/sites/default/files/development_of... (pdf)

The PDF has a complete breakdown of how it works (and I mean really complete - like "here's all the gates in the magnetometers")

You can see some of the interface in https://highways.dot.gov/public-roads/janfeb-2001/safe-plowi...

> This ASP Human-Machine Interface is mounted on the windshield. Some of the worst visibility conditions on the planet can be overcome with the aid of the ASP cab-mounted display, which receives its data from the plow's collision-warning system. Radar sensing assists human sight to find the road and obstacles that may be in the way of the plow.

---

Late addition...

https://conservancy.umn.edu/bitstream/handle/11299/901/20031... has a better view of the interface in figure 2.2

The MN version is GPS based while the CA version was magnet based.

> California suffers from extremely wet and heavy snowfalls in their mountainous terrain. Snowfalls of 4 to 5 feet during a single storm are not uncommon. This wet, heavy snow forces snowplows to operate at relatively low speeds. Because of these conditions and the previous experience the California team (CalTrans, University of California at Davis, and the University of California at Berkeley’s PATH program) has with magnetic based vehicle guidance, the California team focused on the application of magnetic vehicle guidance to this snow removal problem. In addition to the application of the magnetic technology, the California team developed driver displays designed to convey information provided by the magnetic lane guidance system to the driver.

> In contrast to California, issues with snow removal in Minnesota arise because of high winds which can blow light, dry snow at high rates across vast stretches of prairie. This blown snow can create significant drifts which must be continuously cleared in order to keep roads open. Under these conditions, snowplow operations are required to run at relatively high speeds to avoid road closures because of drifting. Because of these weather and geographic conditions, the Minnesota team focused on DGPS based solutions to this particular snow removal problem. The Minnesota team also developed driver displays designed specifically to convey lane keeping and collision avoidance information provided by the on-board systems.


> Road was drivable, but I couldn’t see road markings.

That's a bit contradictory. If you can't see the road markings the road is not drivable.


In Ontario and it's very common to be driving on packed snow for a few days after new snowfall. You get barely any road markings at all. You just use common sense and memory to try and act like the lane dividers, turning lanes, stop lines are still there. It must be the same in many parts of the US as well. With winter tires and the right attitude its perfectly fine!

But yeah if you can't see the road markings due to an active blizzard, flip the hazards on, follow the tail lights ahead of ya, and stay in the middle of were the road should be.


I've lived there. St. Josephs Island to be specific. I also went to the funeral of more than one person that 'didn't make it' through that kind of condition.


> You just use common sense and memory to try and act like the lane dividers, turning lanes, stop lines are still there.

You’re absolutely right. You just do what you need to do.

Having some form of inertial navigation supplemented by GPS and visual odometery from features that stick out from the side of the road (like a road sign; memorized by the vehicle on past drives) would help a lot. A HUD UI could show road markings on “slice” of the road.

Call it “low viz aid” or something catchy.

In my case, I was driving on a rural road just before dusk without street lighting where there is nearly no traffic on a good day. Low-beam headlights were reflecting off the flakes and high-beams were obviously out of the question.

Risk of a collision with another vehicle was low, but going into the ditch was possible if you didn’t know the bends of the road.

Yes, these are dangerous conditions, but I started off driving in rain and didn’t anticipate the switch to snow mid-drive.


But what about a curvy 2-lane road with oncoming traffic on the other side? How do you ensure you not hit kerb the car if you can't see the kerb due to snow?


You slow down enough that you can confidently maneuver by the curb and the other vehicle.


> You just use common sense and memory to try and act like the lane dividers, turning lanes, stop lines are still there

Isn't this what road signs are for?


Here in Sweden, even the road signs are frequently covered by snow. I’m sure it’s a similar situation in Canada!


Interesting. In Russia roads are entirely covered by thick layer of hardened snow for 4+ months (intercity often more, especially up north) so I'm not even sure markings are normative but signs have to be visible. They tell you changes in the number of lanes and directions etc (something like https://www.drom.ru/pdd/pdd/sign_5_15_2/). At first it was unusual to see such reliance on road markings in other countries


Try driving on non major roads (and some major) roads in the UK. They often have no markings because they've worn away!


As a Kiwi moved the the UK, I &%$ hate the roads here.

They're fun to drive on still, but there's an over-reliance on painted road markings, which are often not repainted, if they're even visible they disappear when it's raining, or just you know, when there's traffic and the car in front of me is covering it.

It's the absolute fucking worst on gyratorys and other complex roundabouts where I have no idea which fucking lane to be in because I can't see the road marking, covered by cars, positioned right at the entrance to the roundabout rather than before.

Would be completely solved by ensuring roundabout signs were so far in front of the roundabout depending on road speed, then on the roundabout sign indicating at each exit the # of which lanes you can take to get there, ie mark exits going around with 1 2,3 3,4 5. Then I know immediately that the left most lane is dedicated to the first exit, to take the 3rd exit I can be in the 3rd or 4th most lane. And "going around" has a dedicated lane, the 5th from the left.

So sick of being forced out the wrong exit because I've picked the wrong lane and there's too much traffic. So sick of locals in areas complaining on forums about the "idiot drivers" not magically knowing the road layout beforehand. So sick of having to change lanes and feeling like the asshole in the equation because I know it looks like I've done it to queue jump; I'm not trying to skip ahead, I just have no idea where the fuck I'm going.

Gyratory lane markings are shit, too. Proper lanes and then with each segment no marking at all before the lanes start again, it can often be difficult to tell which lane you need to funnel yourself into after the break because the road numbers on the lanes up ahead just aren't visible until you're right on top of them.

Then the driving test here. Oh, the driving test. The first time I tried it I failed for failing to indicate correctly at a mini roundabout. Okay, fine, fair dues. Pass second time. Been driving a while now & maybe like 30% of people actually indicate on a roundabout. Let alone all the other antics people pull while driving here; pass the strict test & then throw it all out the window.


Those mini roundabouts in the UK can be hilarious. A regular crossing with a black dot in the middle and suddenly it's a roundabout. A non local is utterly without a chance to navigate that in a non heart-stopping way.


Ontario has 10m people and most of us won't see the lane markings for the next few days in the middle of a 4 million metro area. Roads will be eminently driveable and some people will safely use a traditional cruise control.


Not all roads get properly cleaned (esp. non-major ones) in a lot of parts of the word, driving on snow is rather common in North Europe.


Well... there is another problem obviously apart of low traction, which car cannot sense directly. Accumulation of water or snow obstruct the radar. Actually a car with no camera would report exactly that just from radar readout alone, since it can detect worse signal-to-noise due to too much reflection in its readings.


Yeah, the manual explicitly says that cruise control should only be used in clear weather conditions.


Yes, the manual does say not to use adaptive cruise control (Distronic) on icy or slippery roads, if visibility is poor, or if the radar sensors are affected by a variety of weather conditions. Cruise will even automatically disengage if electronic stability intervenes.

I’m mostly annoyed that false positives such as the Nevada issue described above or a one-time splash of water on the sensor when it’s perfectly safe to use cruise control results in the whole system being unavailable for an extended period of time. To me, the right solution would be to disengage Distronic when the sensor can’t see well enough and then allow the driver to re-engage traditional cruise control if they deem it safe to do so. The system doesn’t have the sensors or the smarts to know if it’s actually safe to use cruise control, only the driver can make that judgement. From the manual: “Cruise control is only an aid. You are responsible for the distance to the vehicle in front, for vehicle speed, braking in good time and for staying in your lane.”


There also should always be an option to turn off traction control on gravel roads too. You want the gravel to pile up in front of tires it slows you much faster.

And snow is a pain too. Ford Transit vans were terrible for no manual over ride maybe it's all Ford since I doubt Transit has its own unique system. I used to drive a Transit for work the little goofy looking vans. I couldn't make it through a 12 inch high pile of snow the van kept cutting the engine power. I had to back up hit the gas put it in neutral and coast though the tiny little snow pile. You could pull the fuse to disable the traction control but it also shuts off ABS and the airbag. Useless piece of crap van.


Our Ford cars have a little button which disables all stability program functions, except ABS.

Strange that the van doesn't have it.


Eh, the risk is really overblown in my opinion.

Hydroplaning/sliding is a real risk, but you can mitigate that by simply understanding how your vehicle behaves and paying attention. Sudden, unexpected change in engine RPM, you should probably cancel cruise control. Weird torquing on the steering wheel, disengage cruise and steer by hand.


> Cruise will also disengage in heavy rain or snow, with the system saying the sensor is dirty. It wouldn’t be so bad if traditional cruise control would keep working without the active braking assist, etc. but unfortunately it’s all or nothing.

A previous car I had, a 2019 CRV, did this too. I understand that it could be a very dangerous situation if the driver forgets or doesn't realize they're only using the dumb cruise control instead of the adaptive cruise control they're used to.


> Will I really be able to get replacement parts for the Sprinter when the screen or computer start to die in 10-15 years?

Most likely, yes. Part of why most automotive tech lags behind the times on raw performance is guarantees around things like how long you can source a given SoC, for example. 10 years is not an uncommon manufacturing lifetime in the space, and a Sprinter is a relatively high volume car so I wouldn't be too concerned about MB producing/stocking spares


I wouldn't count on it. I work for John Deere, and am often in discussions about some part going out of production. Sometimes we guess how many we need for the next X years and buy that many, othertimes we rewrite software for a new part. Care to guess how many boards can sit in a warehouse for many years and still have good capacitors.

I can't tell you how many years, but there is always someone in these discussions who points out with pride that we still tractors from the 1950 in use by customers who demand replacement parts for anything that breaks.


In case anyone here weren't a farmer, John Deer is the Apple Tesla of industrial equipment with not just built-in planned obsolescence but built-in inability to self-repair or self-diagnose without proprietary equipment. Much hay has been made in R2R about John Deer's customer-hostile practices.

The harvest time sensor reset tax is why some small-time farmers stick to old and overseas equipment they can repair without a tech visit. It's not about the customer being unreasonable about NLA parts, but the modern unreasonableness of how the manufacturer impedes the customer with artificial scarcity and artificial limitations on knowledge, diagrams, diagnostic equipment, diagnostic equipment manuals, and the nature of some repair parts.


Exactly: farm equipment is one place where you really do not want to Buy American.


> Care to guess how many boards can sit in a warehouse for many years and still have good capacitors.

If you guys are having trouble storing boards for 10+ years with caps regularly going bad, I would highly recommend you look into a new supplier. Well-built caps should have no problem going for 25+ years in storage (especially cold).

In addition, this is probably the easiest component on the board to replace. If you can’t find functional new stock and this is the only issue, just recap them.


I have no insight into how we store things. I sometimes have input into the buy or port to a replacement decision, but if we decide to buy there is a completely different department that handles the details.

Caps are an issue that is easiest for people here to understand, but we have a long list of similar issues to watch for in storing parts


Or ditch electrolytic and tantalum caps wherever possible because they're crap. It's often possible to engineer a more expensive, reliable LR circuit for the impedance of a cheap RC one.


Tantalum have their own issues. Yes, they live longer. But they're nice little fire starters and are much more critical than electrolytic capacitors when it comes to voltage tolerance.


It's kind of amazing to me that we can build 3nm chips but we still haven't figured out how to build a uF-range cap that has long MTBF, good overvoltage tolerance, low series resistance, and doesn't blow up.


Indeed. But... caps are improving, but slowly. The size difference between say a 100 uf 25V cap today compared to one from the 80's is considerable, and if you add another decade it is hard to believe it is the same component.

By contrast inductors have not changed in size at all and resistors are very much limited by power dissipation.


I work on automotive HMI and don't have to guess how many can sit in a warehouse for many years and still have good caps.

They're subject to a lot worse than that for design verification. And I assure you John Deere has many _many_ PCBs that have sat for 10 years to enable service needs today, it's not like they just started using complex circuit boards yesterday...


> built-in nav system is useful for those times need directions and don't have a cell signal.

You know you can also download Offline maps. One of Google Maps best features.


This is one of those things that while true and useful, I personally don't realize or think of it until just after I need it. It's always frustrating to get to a destination without signal and need to plot the return trip, or try to make adjustment in the middle of a dead zone and realize you don't have maps.

I keep an atlas in my back seat just to be safe now.


The nice thing about the built-in nav is that it always works everywhere in the U.S. Google Maps only lets you download an area about half the size of Washington State at a time and you have to manually select the area to download. Fine for local use, but too tedious (and too much disk space) for a long road trip.


?? The Google Maps I use offers to download the maps if I ask for directions on a long journey. It doesn't seem to have download limits and I never need to think about it again.


Have you heard of Open Street Maps?

You can keep offline all the maps you want.


We had a young guy writing good software for us while he still was going to school. We still use it to get things done in production. Then he went to university. He came back for a summer job. He had the sudden urge to ditch everything he used to do and to use the newest unproven technology. Also he had a serious urge to install everything from Microsoft.

Then he created a piece of slow software full of flaws and performance problems - with a lot of unnecessary fluff in the UI. (At some time he mentioned that he was going to use multi-threading for every button, to reduce the lag when drawing the UI).

Then he joined Mercedes-Benz... And makes a lot of money there. And I'm really happy that I don't have to worry about him anymore. I read some job descriptions for Mercedes-Benz and I think he most likely totally fits in there.

What they do to young people in Universities today must be really terrible...


MB is also famous for running Kubernetes. At a 1000 individual setups (all automated of course!). I had the pleasure of flamewaring someone who was very proud of this and couldn't really explain to me, why they would not just get two big physical machines with Corosync for like 70% of those (internal) applications. Doesn't seem much worse than a 3-node Kubernetes cluster (which is going to experience serious disruption too if anything goes down if it is running at full load......) and I doubt that scaling is an important concern with this setup.


At my last company we used k8s not primarily because of the scaling capabilities, but as a way to safely deploy applications.


I've seen it used this way too.

But that's like buying the mcdonalds supersize family meal because you want the fries. There's better ways to go about it.


At my previous job we used docker swarm for a cluster of a dozen machines. Does not have all the fancy features of k8s but gets the job done to automate deployments


It might not be ideal but getting all of that admin and upgrade tooling for free is often worth it


Interesting story. Thanks for sharing.

Not trying to blame anyone here. Did no one guide him? Just by virtue of him being a young student, surely his unproductive proclivities were easily correctable? Was he averse to mentoring?


He already did great things in the past. I liked him a lot before university.

There is too much self-esteem in those young guys. And they blindly trust every new framework that shows up on the market. Also they have excellent grades in university and get offered a lot of money. Somehow guys from management and HR like them. For me there was not much chance to mentor anything.

I could talk more about young guys being problematic... but to me this was extraordinary and somehow related to the things people complain about Mercedes software.

For me two things seem to be dysfunctional: 1) University today and in this case the University closest to Mercedes headquarter 2) HR (as usual...). Things like "practical experience with multi-threaded UI performance improvements" in a CV somehow clicks with HR. At the moment I have to deal with a guy who dislikes databases and instead of adding a column to a table wants the database-server internally to query his webservice when someone selects something from the table. I bet Mercedes HR will love this when he writes a nice line about that in his CV...


> At the moment I have to deal with a guy who dislikes databases and instead of adding a column to a table wants the database-server internally to query his webservice when someone selects something from the table.

micro-column-service oriented schema. He'll go far /g

> I could talk more about young guys being problematic

It's the false sense of 'knowledge' afforded by trivial access to technical content without the benefit of practical experience. It's the internet, not the young men. LLMs are going to max this to an unpleasant extreme.


> multi-threading for every button, to reduce the lag when drawing the UI

Maybe it wasn't needed for those internal apps, but for consumer electronics I wish everyone reduced input feedback lag.


I’m also horrified by the trend to put more and more of the meters into the UI. My Ford has done the booting thing before on a highway as well as once it flipped the screen horizontally until it was shut down and started again. I can’t imagine the quality is any better for the display components that will replace speedometers and such


The media head-end software tends to be much less robust than the digital dash display software. I've had the head unit crash and act poorly on many cars, while digital dashes boot quickly and tend to be pretty solid.

The only time I had an issue w/a digital dash was in a Volvo that had a battery that was nearly flat. The screen itself would glitch in various ways, but the data itself was still solid.


My (admittedly-beloved) 2015 F-150 will reboot Sync if I'm listening to music (usually on Soundcloud) if the song title includes non-ASCII characters.


I'm reminded of: ( https://99percentinvisible.org/episode/the-roman-mars-mazda-... ) the Mazda head unit being allergic to Roman Mars.


Or Janet Jackson's song Rhythm Nation crashing some old laptop hard drives, due to the hard drives' resonant frequency. The manufacturer worked around the problem by adding a custom filter in the audio pipeline that detected and removed the offending frequencies during audio playback.

https://devblogs.microsoft.com/oldnewthing/20220920-00/?p=10...


The concept of technology being unable to handle some input as an "allergy" for some reason makes me giggle.


Awesome, a real world example of the phonograph-destroying record from the book "Gödel, Escher, Bach" ! What a time to be alive! https://en.wikipedia.org/wiki/G%C3%B6del,_Escher,_Bach


My 2011 Ford Focus with Sync would randomly lock up and loop the same 10th of a second of the current track. Like a really old PC if it crashed while playing a sound.

Turning the car and back on again to stop the noise got old really fast. All of the radio inputs stopped working when this happened.


Hey, ring buffer... I want you to play this, and I'm going to give you more data, just keep looping the buffer.

Oops, I forgot to give you more data... or tell you to stop... >:)


Car software is a lot more involved than most of the people here would think. These days, the dashboard display is typically rendered by separate pieces of software. A pretty normal, moderately fancy rendering engine does all the animations and backgrounds while the really important information (e.g. engine/brake system malfunction) is overlaid from a separate, isolated software system with a stack that meets quite high functional safety standards. That is, the chance of the second part of the system ever failing is comparable to winning the lottery.


I don't think I've ever used software provided by an automaker that I would classify as "good". Just varying degrees from "functional, but barebones and clunky" down to "so buggy and useless it's a legitimate safety hazard".

As a result, I'm firmly on team "make every infotainment system a dumb terminal for your phone" and refuse to buy any car that doesn't support both Android Auto and Apple CarPlay.


I find myself torn on this:

* Traditional Automakers: Mostly as you describe, though honestly I am fond of the UI in my VW. Most others score B-D in responsiveness and anywhere from B-F in my "is this UI closer to ideal, or to a caricature of a baffling UI that no one will ever like to use?" scale.

* Apple: "No one needs physical controls for anything, and people HATE seeing any details, so let's try to get as minimal as possible because that's just elegant. Maybe just a dashboard with a your speed, as a single number, unlabeled because labels are ugly, and three dots (touch target 0.6cm) that you have to touch to access a menu to access all other features."

* Recent Automakers: "People love screens. Let's delete lots of switches to save a few dozen dollars, and stick functionality in deeply-nested menus that require you to take your eyes off the road."

I really want the $20 worth of physical buttons, knobs, stalks and switches of a traditional car, and the extensibility of the CarPlay world.


A colleague with a recent Land Rover Discovery Sport has screens on the buttons/dials. Little mini screens for things like an AC knob.

It’s a Land Rover, so obviously those screen crash/bug out then he gets no HVAC.

He hates that car. It’s beautiful, luxurious and a pleasure to drive. But it’s been in the shop >50% of his lease.


I can't find that comic where the head unit is just a rectangle with a 3.5mm jack. That's all we need, really.

My old car has a cheap phone clip screwed into the dashboard, and the phone connects to the speakers for music. I can't imagine anything improving upon that.


CarPlay is definitely a huge improvement on that.


I feel like a fanboy, but I absolutely love my Hyundai Palisade. It's legitimately changed driving for me. Some quirks and annoyances, but absolutely nothing as bad as listed here.

* Software is simple, consistent, and functional. Some goofy choices, but they're mostly visual preference than meaningful usage degradation.

* Hardware buttons support all of the core UX.

* Lane centering and adaptive cruise control as rock solid. I've legitimately had the car follow lane lines that I was struggling to see (rainy with reflections).


Tesla is the only one I'd classify as good enough for daily use without needing my phone. Possibly because they started as a tech company so they understand software.


Most automobile engines made in the last 30 years are running software that works at least okay.


That sw does not come from Mercedes, they just buy different devices for what they need from suppliers. The salary offered by one such supplier for a graduate engineer was less than what a construction company was offering to hang dry wall (qualified worker). The construction company was raising a new building for the supplier.


That's mostly an irrelevant detail.

Mercedes may not be the OEM in this instance, though they are certainly the vendor, who put their name and reputation to it.


Exactly, the buck stops with Mercedes. It's on Mercedes for not investing in higher quality software for their vehicles.


That was their choice though.


Some Mercedes also have the bug where the sign classifiers somewhat regularly mis-read 80 as 60 for speed limits, causing cruise control to arbitrarily slam on the brakes in some parts of the country. You would think that they would catch these things with rudimentary quality control.

The automotive code for many of these systems is incomprehensibly awful but the entire process of how it is made virtually guarantees this outcome.


I just ran into this issue in Montana the other day. Weird thing was that it’d sometimes read the 80 mph limit correctly, then a second or two later revert to 60. My Sprinter just beeps at me, though. No brake action if it thinks I’m speeding.


Everyone I know that has discovered this bug, discovered it in Montana. All of the States that have ubiquitous 80 mph speed limits are clustered around that area.

It reportedly happens with 80 kph speed limit signs in Canada as well.


Texas has an 80 mph limit on a long stretch of I-10. They have one other highway (130) with an 85 mph limit.


Wonder if the “80” is from maps, which is then overridden by buggy vision?


80 mp/h speed limit?!?


I just borrowed a Tesla Model 3 that stopped for green lights. I asked a service person: apparently it is known to stop for a green light if there is no car ahead. The display shows a green light, a red line, and fine print saying that the car will stop for traffic control.


I don't understand this. What does stop for a green light mean?


It treats a green light like a stop sign


Great way to get rear ended.


The other thing that worries me is how integrated modern headunits are. It used to be that when your headunit broke or became obsolete, you could just whack a new one in for a few hundred dollars, prolonging the life of your vehicle. Now even basic cars have fully integrated headunits making upgrades impossible. Even if the headunit continues to function, the technology is going to become so obsolete that they’ll be virtually useless in 20 years. I doubt the iPhone 35 will connect to a 2023 car. I guess people who buy new cars couldn’t care less about what happens to them in 20 years, which is an easy win for car manufacturers looking for ways to reduce the effective lifespan of their vehicles.


Yep, exactly. My personal vehicle is a 2000 model year Toyota. It has an incredibly good, stable, and modern CarPlay head unit from Sony, because back whin the car was made we had standards like double-DIN head units and you could just plop in a new stereo. Which is exactly what I did, and now I have a "golden age" mechanically reliable, nigh-un-killable Toyota with a modern carplay head unit.

And as a bonus, I can adjust everything about the car (climate control, volume, overdrive, 4wd, ECT mode, etc) with real buttons and tactile feedback while going down a washboard track, without ever having to take my eyes off the road.

If anyone out there starts making EVs with the same bare-bones attitude (analog knobs and dials, no flashy infotainment system, double-DIN head unit slot, modern safety features and a simple ultra-reliable EV powertrain), sign me up. It'll probably never happen because there's not really an economic incentive right now to try to hit lower price-points in the EV market, everyone's going upmarket. But I'd still love to see somebody try.


>couldn't always be solved by power cycling the vehicle.

From experience, I've found that turning a car off is akin to putting a computer to sleep. If the state is fubar, no amount of turning the car on and off will fix it.

The only proper way to "power cycle" a car is to momentarily disconnect the car battery (remember to turn the car off first) so all power to the computers are physically cut off and they /have/ to start from a cold boot next time you turn the car on.


Disconnecting the battery can have other consequences. Easier to reset the fuse


A good idea, put a fuse + switch there, and add it to the dash. 1/2 an hour of work.

Better yet, do so, then demand part costs + time + debug + this is the only way to fix it. A $1000 seems abiut right. Having the dealer fail to fix it (all those reflashes) helps, for the small claims court case, as surely they will refuse to pay.

Then post the court case win online.

This seems to be the only way to get big anything to do right by the consumer, these days.


Contrary to all the legends of "German quality/engineering" (of which there is certainly in certain industries) I've heard that German cars were shit during the 90s.

Japan OTOH seems to be very aware of how these newfangled tech can screw your reputation.


Honestly, I feel like German cars have an undeserved reputation for quality. German cars are superficially well built, quite fast, and make satisfying thunks when you close the doors. However, they are fairly unreliable, sensitive, and when they do fail, cost a lot to repair.

I'll take a Japanese car any day, however, Hyundai, Volvo, Skoda, Kia, Peugeot and Ford are also better engineered than most German cars nowadays in my opinion.


This is just a personal anecdote - I got a brand new Mercedes in 2016, it was even fully made in Germany - and it was probably the worst assembled vehicle I have ever owned in my life(and I used to own a 1995 Fiat Cinquecento), the number of creaks and rattles coming from all places in that car was almost funny if it wasn't so upsetting. I've had several visits to the dealership just to fix rattling seats, upholstery, dashboard and sunroof. Drove very well and never had any mechanical or electrical problems, but the interior was horrendous.

Then in 2020 I swapped it for a brand new Volvo XC60(made in China!) and in the last 3 years this car has literally been completely trouble free. No rattles from anywhere, nothing. Extremely well put together, comfortable, drove it across Europe multiple times now and literally no issues with it whatsoever. Such a stark contrast to mercedes for me.


Are you sure it's made in China? For most markets XC60 are built either in Sweden or indeed in China, but I always wondered about chinese-produced ones. Fellow Volvo owner (XC90/2022 - Sweden-built though), car's a dream.


Yes I'm very sure. The car was built in Chengdou in January 2020 then transported over here by train (option which no longer exists at the moment because Russia closed the land link so cars have to be brought to Europe by ship the old fashioned way). It's a T8 plug-in hybrid for the British market.


Mercedes' reputation for reliability is mostly a legacy of the 60s and 70s. Their cars from that era were absurdly overbuilt, especially the diesel powered models. You will still find diesel MBs from the 70s on the road today, some of them having racked up 500k miles or more.


My father has a 98 S320, with over 1 million kilometers. The engine just recently started to have overheating problems and needs to be rebuilt with new gaskets, but otherwise runs fine!


As a former Audi owner, the old saying was absolutely true: if you can afford to buy two Audis, you can afford to own one. Absolutely fantastic vehicle until it blows up and then you are buying VW parts with Porsche price tags.


Where does this German engineering saying come from? I find German engineering unnecessarily complicated and it looks more or less designed by a committee of bickering members.

Japanese engineering on the other hand in my experience is very solid and also pretty cheap. It is easier to repair and vehicles are like mountain goats that can go anywhere and can be handled roughly.


Skodas are just previous gen Volkswagens btw


I understood Skodas are usually _next_ gen Volkswagens. At least when I lived in Europe they always got the new chassis/engine/gearbox to beta before it rolled out to VW, did that change?


My Skoda is the most boring, unsurprising, uninteresting and reliable car I every driven. I love it.


I'm sad they've stopped making functional-first cars like the Yeti, and now just do versions of VW models. I think Dacia might have taken up the mantle of no-frills and reliable though.


Not even previous gen any more. They just get interiors that aren’t quite as nice (more plastic).


I've driven mostly BMW's and their build quality has been decent.


The 90s was a sweet spot for the three luxury German brands: Audi, BMW and Mercedes Benz. Mercedes in particular has had a very marked decline in quality since then.


Like what’s the highest quality luxury brand now, Lexus? Given that it’s Toyota engineering behind it I’d assume they are still good but I’ve never been in that market so idk.


I am currently renting and driving a Tesla from Hertz at SEA on mainland for a week of skiing at Mt Baker. I have a 2020 Lexus RXL at home. While the Tesla has better driving software and sportier handling, the Lexus rides much better and quieter and just seems nicer. Anyway, love the Lexus fwiw.


Still those three, plus Lexus, Infiniti, Acura, Range Rover etc.


Are you claiming that Land Rover makes a high quality product? Or just a luxurious one?


Mostly a luxurious one today. I test drove a 2012 Range Rover in 2020 and all the rubber on the controls had melted. Would Not Recommend, but they are popular certainly.


RR is overpriced junk.


I had a 2001 Audi and it was by far the best vehicle I've ever owned. I bought a 2016 Audi trying to replicate that and I was dumbfounded by how little it had changed in 15 years, but what _had_ changed seemed to be distinctly for the worse. I won't be buying another one :(


> Japan OTOH seems to be very aware of how these newfangled tech can screw your reputation.

My toyota is absolute crap, if this helps.


What's wrong with it? But yes it seems like more recenr toyotas are crap, especially the interior.


It has a neat feature of auto locking itself if you forgot it, so if you forgot your keys in, they will be autolocked in.

The positions of the pedals is not fit for humans (but that's a bit OT)

The integrated not readily replaceable without servicing the entire dashboard USB port just broke.

The feature to alert me when I'm about to hit an obstacle triggers when I've already done a full stop.

The radio touch interface is awful. There is basically no way to say "play everything". You've got to select the artist, then the album, blabla. Of course you need to stop the car to move on to the next album.

The detection for speed signals is so wrong. It consistently gets the signs that are meant for slow-down lanes to exit the highway as the actual limit I'm supposed to have.

In some specific places it imagines ghost speed signs every single time I pass.


Apart from the electronic junk how is the actual driving experience? Can you drive comfortably with good power?


> Can you drive comfortably

As I said, the pedals are placed in a position that was not designed with human beings in mind. So, not really. Quite uncomfortable.


someone should let toyota know, because their software sucks 5 kinds of taint on my 2020 taco


Mercedes' crap software caused me to stop driving them, even though I love the handling, the reliability of the engine and the general lack of distractions. They really blew it, their software is a safety liability rather than an asset and I really wonder what caused them to drop the ball in such a terrible way.


Do you know any programmers who think it's cool to work for German car companies or that they at least offer a decent pay? I think that's the answer: those traditional German industries don't care nor understand IT. These days and entry level KIA has better software than a BMW (I owned several BMWs from different eras and will not buy anything from them again).


BMW is just plain embarrassing. A long time colleague was a huge fan of the brand, he'd buy pretty much all of their top of the line models one after the other. Some of them were so often in the shop with software issues that he ended up getting attached to the replacement vehicles :)


> couldn't always be solved by power cycling the vehicle.

Made me remember when I turned on my new (dumb) TV and it did a startup sound. I knew it would get worse in the future.


I really despise smart TVs. My new Roku tv poops itself all the time and the apps stop working so I have to look for the system reset and it's not always super straight forward.


This is why I only buy used MBs from pre-2016. They're fantastic and they don't require a subscription. Though the 2016 E/S is a wifi hotspot if you really want that in a car.


> It had obvious race-condition type bugs when it came to the user interface layer, but most frustrating was its tendency to succumb to some kind of memory leak on long drives where the entire head unit would just lock up and crash to a black screen after 6-7 hours of being turned on. Because the vehicle kept the computer system "warm" for up to 30 minutes or so to avoid doing a full (and slow) bootup process every time you stopped for fuel, this was a real problem on long trips and couldn't always be solved by power cycling the vehicle.

Had this problem with an A-K200 about 13 years ago - when driving long distances, after a couple of hours the nav system would randomly stop working. Sometimes stopping at a gas station would allow the system to reset by itself, sometimes not.


This comment is extremely enlightening. I'm really considering a Sprinter van as my next vehicle purchase.

We do a _lot_ of driving on rural roads where we can count on a single hand the number of cars we pass in an hour.


I own a MB V-class and had no software issue I remember.


Beware of Sprinters: they contain Dodgy parts.


Your final conclusion surprised me. All of the software woes you describe sound like they come from insufficient foundational technology, so why not be pleased that they are now investing in those foundations?


German Software Engineer here. I'd rather chop off my left pinky than working on any software product any German car manufacturer is developing:

1) Their management hierarchies only consist of MBA suits that think short term and treat their engineers like apes in a zoo. This frequently leads to engineers actually behaving like apes in a zoo.

2) German culture is deeply based on thinking in silos. They divide everything into boxes in order to organize it "the right way" and don't understand network effects that occur if you focus on building a great (software engineering) culture. Instead, they mindlessly copy-paste whatever fancy thing they see in a silicon valley context.

3) My hypothesis of why German IT products are almost always horrible is that they're rarely thinking from customer's (or developer's!) perspective. They may be proud of some specs of their products but the overall thing is then often just a bunch of related subsystems "pragmatically" tied up with some duct tape.

They still obviously don't understand why they're being outsmarted by Apple, Amazon, Google and others. They're aggressively arrogant and rather lobby the German gov't or cheat to achieve short-time success instead of reorganizing their structures to still be leading in 10 years.

I bet, they internally expect industry leading products 2 years after launch and then will quietly kill the whole thing when they realize that their bet was insane from the beginning.


I lived 9 years in Germany, now in Poland. For me the fact that in Poland a programmer earns significantly more than a Financial Controller was an initial shock a few years ago. But now I understand the problem is with Germany’s stubborn attitude towards everything IT, IT is just considered easy and not “real engineering” and a thereby a cost to reduce. Financial Controllers pushing numbers in Excel files are adding a lot more value, apparently in ways I still don’t fully understand. Honestly, with those salaries and that culture Mercedes is probably not getting the best talent out there.


> 3) My hypothesis of why German IT products are almost always horrible is that they're rarely thinking from customer's (or developer's!) perspective. They may be proud of some specs of their products but the overall thing is then often just a bunch of related subsystems "pragmatically" tied up with some duct tape.

While that is commonly true, I've also found that driving a modern Audi or a Mercedes is still much better thought out from UX and comfort perspective in comparison to many vehicular products from Sillicon Valley collaborators. Software might be a bit laggy, but many times there are plenty of nice user respecting features that you won't find in modern agile products (basic operations like "reroute me around this closed road" or opening a glove compartment are stupidly complicated or impossible on brilliantly developed products like Tesla).

The Germans also have one other benefit: my car UI doesn't just randomly completely change with updates like Android Auto or Tesla UI does. I'm not a guniea pig for some product manager trying to get his promotion with redesigns.

There has to be a balance somewhere between the dinosaur process of German brands and coked up 5-second attention spam of Sillicon Valley products.


A friend is working for Mercedes research. He can confirm all your points.


I also worked in the German car SW industry and can confirm what he said. They just don't value SW development. The only way to make decent money there was to become a manager and climb the corporate ladder. Programming was never valued and always under the pressure of off-shoring or throwing the work to talent on a visa from abroad who won't have high expectations. The closer you were to the money, like financial controller as said by the sibling, or upper management, the more you were valued, while the closer you were to "the trenches" the less you were valued.


Absolutely the truth has been spoken! In my experience the management are physics or mechanics engineers. The rest, I can only confirm. I would buy any day of the week a mechanical thing from Germany tool, or whatever. But when it comes to software NO F*ING WAY! Absolut trash.

For more details, see my other posts on the topic: https://news.ycombinator.com/item?id=34164748 https://news.ycombinator.com/item?id=34901394


I've been working a few years in various SW-related roles in the broader MB ecosystem. It helps to understand those german car manufacturers are not monolithic entities, there is a confusing system of subsidiaries dealing with software, often overlapping in scope and competing over projects. Hence, a MB software engineer would often work at DFS, DMS, MBition, FreeNow or one of those shops. By design those do not adhere to corporate standards and they emulate start-ups, i.e. k8s, aws, slack, table tennis, bro culture, beer.

It's a curious setup because most of those subsidiaries ("corporate startups") are huge money sinks and produce little of value. Salary is decent/good, the offices are lavish, and the tech can be interesting. The downside: you cannot shake the feeling of being on a large playground with rich, generous but eventually indifferent parents. I'd say in germany, it's a good place for juniors to pick up some technologies and skills, quite a few former colleagues ended up at prestigious tech companies later.

Core technology like autonomous driving, Engines, CAN, etc... is developed in-house with more rigor, compliance and process, and the disregard for "software engineering" is also due to those agile software subsidiaries producing sloppy 2nd tier, unsuccessful auxiliary technology like apps, services or entertainment stuff. Unlike maybe Tesla, MB or BMW could not afford the reputational damage for selling half-broken autonomous driving.


It is not just Germany, but software development culture everywhere else that is not Silicon Valley tends to be like you describe. There has not been great Nokia software, great Samsung software, great Sony software. Only hardware. This is why challenger banks and banks app are so popular in Europe: they have a UX that works - something that old organisations constantly fail to produce.

Cars are just one piece in this greater cultural phenomenom. There is a reason why e.g. Android Auto is better, with all its flaws, than the competition. There is a reason why people like Tesla's software.


"chop off left pinky" - that's an interesting thing to say. Have You by any chance been working for a small Japanese outfit these past few years?


No it's a shout-out directed toward all the emacs users out there


Yuck:

> MB.OS is designed and developed in-house to retain full control

You better have some world class software engineers to pull this off, a LOT of them. And retain them. If not, you better go open source. You can still retain full control with open source.

Good software is the KEY to good cars, especially electric ones. Why are we inventing 100 different models of cars every year? What a waste of human potential.


Ecosystems are the key, not so much in which house or how public it is built. (see: iOS) It's also a system for a car, and probably a niche one at that.

There are a few things that are relevant in Car-OS these days:

  - Ecosystems
  - Long-term maintenance
  - Integration with a phone that takes over most day-to-day interactions
  - Car should work without it
If MB.OS and BMW's version and Audi and whoever else wants to build one doesn't support CarPlay or Android Auto (or whatever their future incarnations will be), it is already irrelevant. If anything, MB.OS is about as relevant to a consumer as QNX is. Or VxWorks for that matter. (spoiler: consumer doesn't care and doesn't really want anything to do with it)

That doesn't mean that all car makers should just keep using an ancient QNX version statically forever, or whatever some OEM happens to install with the head units, but an in-house Linux distro with some vendor branding sauce really is about as relevant as blinker fluid.


This makes an odd assumption about what drives, and does not drive, sales. Companies don't buy 300 company car because it has the best center console OS.


No, it doesn't. What I wrote is the exact same thing you condensed here:

> Companies don't buy 300 company car because it has the best center console OS.

However, I centred mine more around personal consumers, rather can organisational buyers. It's also not just some console OS, but an ecosystem that also reals with sensors and self-driving features.

I also went more into the technical aspects of operating system success, but rather than say something to the extent of "yet another QNX-style lineage is about as useful as Windows Mobile", I wrote a slightly more expanded story, just like Mercedes-Benz did on their announcement (but they added a whole lot of extra fluff about how their OS is a value add).


> If anything, MB.OS is about as relevant to a consumer as QNX is

QNX isn't exactly defining for an ecosystem, MB.OS probably will be.


I beg to differ with that. If you shut down QNX globally it would pretty much end civilization. So much stuff runs on that little core that it boggles the mind.

The great gift of QNX is that it is so reliable that you forget it is there. But no mistake: it is still there and it still works and without it you'd miss it right away.


Yep QNX and VxWorks, for smaller stuff RTEMS and FreeRTOS are just everywhere.


I’d be surprised if this isn’t just another Android fork.


I work there, and I'd be massively surprised if it were.

Historically, their in-vehicle software has been handled by other companies they've contracted, and is radically different from generation to generation (different company each time).


It's not from what I read. MBUX, the UI is built using Qt.


That kind of implies that the new Mercedes OS will have some kind of POSIX API, since Qt doesn't really work without it. Their new Qt/MCU bullshit doesn't count.


Windows isn't POSIX, and Qt does just fine there.


Sorry, was talking more about embedded systems bindings. Qt has shims, obviously, to work with Mac/Win/GTK etc but when you're setting up Qt as the windowing system, it needs POSIX.


I'd be pretty sure that any company can pay money to Qt Company and have them come and integrate Qt into any kind of weirdo OS whatsover.


Exactly like Tesla.


I disagree. Sometimes you need to invent things yourself. Mercedes is an absolutely massive company with many subsidiary brands. This is not an automatic bad move.


But they are not a software company at the core. They need to be to pull this off.


Both Mercedes and Volkswagen have absolutely massive software divisions, easily 10k+ people. They pay really well for German standards and have access to a large talent pool of people that would like to live in Germany.


Big and well paid doesn’t equal quality. The fiasco of id3 is the best example. Or the whole Audi/VW drama. It doesn’t change the fact that their management just doesn’t get software. Pouring more money over that and hiring more “talent” isn’t going to change that.


Most hardware companies are not known for good software. They treat software like it's just another line item on the BOM: Like a bolt or a screw. Procure it as cheaply as possible, scoop it into the product on the assembly line, and make sure it passes a few checkboxes and you're done. Unless they see software as a core differentiating product feature that drives purchases, their quality bar is going to be "barely good enough to not cause a product recall".


These days even most software companies ain't known for good software :D

There's very few quality software products these days. And that's usually just a temporary state before eventually a product gets messed with :(


> The fiasco of id3

Could you explain what specifically you refer to?


https://www.theverge.com/2020/6/11/21288572/volkswagen-id3-e... + general search for "software id3" shows that issues plague this car to this day.

Diess lost his job because of that: https://www.theregister.com/2022/07/26/from_beetles_to_bugs_....


>Both Mercedes and Volkswagen have absolutely massive software divisions, easily 10k+ people.

That doesn't mean jack. Most large German companies I worked at tended to be super inefficient, being very top heavy and prone to constant turf wars and power struggles between the managers, as that was the only career path that made you any decent money so everyone was focusing on that instead of building great products.

>They pay really well for German standards and have access to a large talent pool of people that would like to live in Germany.

Meh, pay is underwhelming, as they focus more on getting off-shore talent from abroad on a visa who's more interested in getting to Germany at any cost rather than netting a higher market wage or increasing wages to attract local talent.

These companies are good places to coast on average wages, not to make money and innovate making great products.


>They pay really well for German standards and have access to a large talent pool of people that would like to live in Germany.

But they don't pay well for IT standards, German standards (IT) are lower (sometimes a lot) then polish ones.

>large talent pool of people that would like to live in Germany.

From Hungary, Romania or Turkey? Because no one else in Europe would like to work in Germany (in the IT industry). A bit of reality salt would be better for the German IT-Industry.


Most of their software is done outside of Germany.

MBRDI, known for its engineering innovations, has grown to a team of over 6,000 employees and is one of the earliest technology and innovation center of a global automotive company to set up a strong presence in India.

https://www.mbrdi.co.in

At Mercedes-Benz Research & Development North America, we develop and certify cars for the U.S. and the world in six locations. Our work ranges from powertrain development and vehicle certification, to autonomous driving technology and in-car user experiences including advanced interior and exterior design. It is not just about cars, it is also about creating the latest and greatest software, cutting-edge technology, and groundbreaking innovation.

Locations:

Ann Arbor, MI Carlsbad, CA Long Beach, CA Redford, MI Seattle, WA Sunnyvale, CA

https://group.mercedes-benz.com/company/north-america/merced...


Volkswagen planned to hire 10000 new people for their "Car.Software" division in Wolfsburg, Ingolstadt, Berlin, as well as Seattle und Peking until 2025. They are also building up their hardware division with plans to do cutting edge node designs in 5nm and beyond. I know that they have extensive recruiting efforts in both areas.


Do you think the talent pool of people that want to live in germany and don’t want to make substantially more working for american companies is large?


Have you lived in Germany? I would only consider going to the US for substantially higher pay. Also there are 80 million Germans, many of whom are fairly well educated and would like to work for Mercedes in Stuttgart.


Oh, I didn’t mean immigration, I meant there are US companies that have offices in Germany that pay more and likely scoop up the local talent.


I mean their self driving tech seems to be the most impressive out there


Umm, no. I tried their Drive Pilot L3 system in Germany over the Christmas holidays (MB EQS) and of 8 hours driving, it was only available for me to drive for 12 minutes (and not a continuous 12 minutes, either). Granted this does not mean the system is poor quality, but it's horribly impractical... if it's rarely to ever available.


It sounds like they are more responsible than the other companies.


Exactly. Using your customers as guinea pig data collectors for an undeveloped system seems to be the go-to. There have been noted fatalities due to this.


What are you basing this assertion on?


Can you elaborate on why they need to be?


The name MB.OS is misleading. It is not a new operating system but a linux or android based hmi sytem.


My bet is that the name is due to internal empire building or hype to shareholders. Being in charge of "MB.OS" sounds better than "another attempt at a good infotainment system". Add in trendy "digital" offices and you've got yourself an aesthetic.


I think it is simpler. It used to be MBUX and that was hard to communicate to people on the street. MB.OS is a lit closer to iOS which helps.


What constitutes an "OS"? It's not entirely unfair for Mercedes to call a customization of Linux an "OS". If it includes GNU coreutils, the linux kernel, but a new init system and communication or UI libraries, you could call that "Mercedes Benz OS". A marketing whiz certainly would!

Or you could even s/linux/QNX/g. But in any case, running a trusted and reliable kernel that wasn't designed by an auto manufacturer could still be what they're doing.


Good cars are the ones that have the least software imo. I’d prefer my car to not have adaptive cruise control, brake or lane assist, etc. More failure points and less control over my own car


> Why are we inventing 100 different models of cars every year? What a waste of human potential.

Human efficiency and human potential are very different things. I don't want to be an efficient cog in a well designed efficient machine. I want to be a human being - building cars is cool and fun. Fun is probably a waste too, I suppose. "Two plus two makes four without my will", etc.


Think of all the cool and fun things that no one has built yet though.


Like all the cool cars that haven’t been built yet. Some people really enjoy them


Those are definitely included


The best car OS so far (even better than Tesla IMO) has been Apple CarPlay. Just be dumb and let my phone handle it, and make the touch screen decent. Bonus points if you implement the multi screen CarPlay concept! Just give up!


This is the same reason I do not want a smart TV. I have one (a Sony), but thankfully it must have lost contact with our WiFi network at some point and I never notice that it does anything other than act as a monitor for our Apple TV. It used to be that it would periodically bug me about software updates and I would endure its dreadful UI just long enough to get through that.


This is the saving grace.

And luckily, the next generation CarPlay will span the entire dashboard, including in front of the driver [1]

I can’t wait to abandon my native “infotainment” cluster**

[1] https://www.macrumors.com/2022/12/28/next-generation-carplay...


It will still require stupid touch screens and will not integrate with custom hardware (HUD anyone?). It is subpar experience now, and using multiple screens to show even more "widgets" is strange value prop. The widgets I am interested while driving are not coming from some app. I want these screens to show information related to the car - cameras and parking sensors, cruise/lane control data, tire pressure, etc. Not a meeting reminder or current time in Tokyo.


The 2 main things I look at on my car screen are "now playing" and navigation, both of those come from third party apps and both would be handy to have situated between the dials so I don't have to look off to the side


CarPlay already doing navigation-in-the-instrument-cluster part in my car. It is much worse than built-in navigation integration. It is just a stack of direction arrows. Now Playing works perfectly fine via regular BT over the last 10+ years. Any rental car I had will show it in the instrument cluster.

Interesting, you did not mention speed or gas/charge gauge as the main things to look at. I for one will not trust apple to display my current speed (see mockups shared above). I've seen plenty of cases when CarPlay will pretend that it is working (play music, show message popups) but totally forgets to update navigation position since 10 miles ago.


Given CarPlay history, either very few OEMs will actually adopt this or it's going to take a long time for your favorite car to get it. Sadly, no one wants to give up control to Apple and Google.


“Apple says the first vehicles with support for the next-generation CarPlay experience will be announced in late 2023, with committed automakers including Acura, Audi, Ford, Honda, Jaguar, Land Rover, Mercedes-Benz, Nissan, Porsche, Volvo, and others”


They had similar statements when CarPlay was first announced. It took years for OEMs to support it. Maybe it will be different this time.


This is a nice idea if don’t plan to have your car for more than 2 years or to never actually own it.


This looks amazing, tbh.


Looks stunning, but no thanks. I want analog dials and physical buttons and knobs for features like gauges, air conditioning, and volume.


Yes! Screens and flush buttons are not good UX for a car. One should be able to change volume, air conditioning settings, radio station/song without taking their eyes off the road.


Mazda is great at this. They don’t pretend to make better software than Apple. They just give you nice hardware interface and a reasonable screen that you don’t have to touch to manipulate it while driving a large chunk of metal very fast.


Agreed, except the last bit. My phone should have nothing to do with the UI showing speed or RPM.


I can get behind this. I'm okay with a simple car display for the vehicles vitals.


Or analog dials and indicators instead of a display


Yes! Every car infotainment system should only cover the basics (bluetooth, radio, critical car controls). Then act as a dumb terminal for CarPlay and Android Auto.

Cars are trying too hard to do things our smartphones already do way better.


Really dislike that idea. Forces all the ios stuff on the dashboard, usually in low quality and limited controls.

BMW CarDrive starts getting bad but it’s still pretty decent good quality, fast os. What irks me about it is that the BMW people start dumbing down the ICE UI to match the simplicity of the EV, which takes away a lot of of good stuff.

Best visible when comparing the 2019 vs 2022 model UI. 2019 was by far the best.


Right. I don’t trust Mercedes to build an OS any more than I trust Apple to build a car. They should focus on their strengths.


That’s not a car OS; it’s the infotainment part of the car’s software.

FTA: “MB.OS benefits from full access to all vehicle domains: infotainment, automated driving, body & comfort, driving & charging”

I don’t think you are arguing your phone should connect to the cameras, acceleration sensors, LIDAR, etc. in your car and handle the self-driving, do you?


I'm hoping thats what the "apple car" project actually is. Mobile devices are getting very, very powerful. I would love for my phone to handle all the processing for my car in the future.


You would love your brakes to depend on a USB or Bluetooth connection?


Anything that doesn't require the processing power can and should still run on the car, brake by wire is totally fine. I just don't mind if my phone has access to a few more sensors here and there, having CarPlay tell me if my tire pressures are low or being able to adjust my EV's profiles from the CarPlay UI would be nice.


> The best car OS so far (even better than Tesla IMO) has been Apple CarPlay

Which in itself is a very interesting statement.

It's taken a while for that to sink in - Apple Computer, who have never designed or manufactured a vehicle, make the best vehicle entertainment system.


Tbh I think Android Auto is better. Both mine and their statements are just conjecture.


JFYI, Apple CarPlay never have been an OS. People often get confused between the foundation SW that OEM provides and consumer overlay that companies like Apple, Google provide. Both are not the same. One cannot replace another.

You can refer to this blog about more popular misconceptions among consumers. https://blogs.blackberry.com/en/2022/10/the-top-9-misconcept...

Disclaimer: Worked for a Teir1 SW provider in past.


This is why I can't wait until the next generation CarPlay is released. When your phone is rendering the infotainment and the driver display it's going to be game changing based on the examples Apple showed.

Android Auto/CarPlay changed car infotainment forever, and the next logical step is to just take over the whole thing.


throwaway account

- the cheapest supplier/contractor gets the contract. Always!

- the salaries in engineering are among the lowest on the market (at least in Germany). The headcount increased from 200 to 800+ in just a couple of years but so did the churn rate and focusing on the salary when hiring, means a lot of talent is skipped from the start (candidates should be happy with 20% less salary given that they work at Mercedes :shrugging Obama meme:)

- (lower) managers and engineers are generally apathetic and disinterested - about the products, their teams and the company. Higher management lives in a parallel universe.

- and maaaaany more issues

I know I'm bitter but, looking at Maslow's hierarchy of needs, how can you expect your employees to reach their full potential when, as an employer, you don't even offer them the basics: a decent salary, support, guidance etc. </rant>


My family had mercs since the beginning of the 90s and its very apparent when those problems started.

I used to say "life behind a star is different" and now they all look the same, feel just as badly engineered as the next VW and have more plastic in them then a modern store for drinks.

Your not bitter, its the truth.


MBition maybe?


bingo


A friend just bought a Mustang and showed it off.

The driver's dashboard is a somewhat configurable screen. All I could think of was that a 1970 Mustang, treated moderately well and maintained regularly, is still a useful car fifty years later. In fact, if it's survived, the asking price is within spitting distance of a new Mustang.

Is there any chance that a 2023 Mustang will be a useful vehicle in 2073?

Will any 2026 Mercedes still be running and maintainable in 2076?


I agree with your general point regarding the durability of older cars. But any 1970 Mustang still running and in good shape has required a lot of maintenance and work. Part of me wonders whether the move to electric vehicles from ICE will resemble the move from CRT to LCD televisions. Less robust and initially more expensive, but high consumer satisfaction and eventually economies of scale will make it cheaper.


> Part of me wonders whether the move to electric vehicles from ICE will resemble the move from CRT to LCD televisions. Less robust…

EVs are far more robust than ICEs from a mechanical level, as the number of moving parts is reduced by orders of magnitude. The problem is electronics, but this has been plaguing ICEs for over 30 years. There’s no reason for the electronics in an EV to any less robust than an ICE, but we are going to see far fewer mechanical failures.

EV or ICE, it seems unlikely any of these cars will be running in 30 years let alone 50. Eventually the electronics fail to a point that is too difficult to diagnose or fix.

Case in point - had a friend with a mid 2000s Astra. Everything worked except for the left indicator. Eventually the car was written off as no one could fix it. The car was just over 10 years old at the time.


Maintenance is cheaper for older cars, and parts are readily available. New cars are much more expensive to maintain, and the cost of parts is an order of magnitude higher.

All of the cars being produced today will be junk in 20 years. Actually, I take that back; they are junk the moment they roll off the line.


> Is there any chance that a 2023 Mustang will be a useful vehicle in 2073?

Do tou mean that EV SUV that some MBA called a Mustang? Do you think that battery pack will be useful 50 years from now?


In fact, my friend bought an ICE Mustang. The judgement was that the EV is overpriced unless the cost of gasoline skyrockets in the next 5-7 years or the price of electricity plummets.


Why does the automobile industry insist on keeping these things in-house when they are clearly terrible at them? I recall an interview with a Ford manager a few years back where they were asked something about adopting Car Play and the response was something like “we have people that can write software for us, so we won’t be using that.”


> The company is confident that this strategic approach to software and hardware development will be the basis for lifetime revenues as well as additional contributions.

I think they're pretty blunt as to "why." What they seem to be hazy on is the "how."


Money.

Why give that up when the customer is really only going to use bluetooth audio anyway? Anyone who cares enough about navigation is gonna use an app on their phone. Native car navigation is a superficial value-feature not a use feature.

The edge that Carplay gets is really minimal. I'm repulsed by the new car market, its many multiples over-connected anyway. When it comes to digital features my gut says only the safety ones drive sales.


> Native car navigation is a superficial value-feature not a use feature

I thought the same thing until I owned an EV. Using the car's navigation is essential for road trips that involve charging along the way. Maybe some day Google Maps et al. will gain access to API's to read the car's current charge and enough information to accurately predict charge along the route, but for now I have completely changed away from using 3rd party nav and rely only on the native vehicle nav.


> Using the car's navigation is essential for road trips that involve charging along the way.

Sounds like a problem that will get solved automatically once charging stations are as common as fuel pumps.


In that case, why even spend money trying? Just deliver cars with room for the customer to add what they want.


Then you don't get to upsell on features and you're less competitive with everyone else.


It seems like they go through phases of we-can-do-this-in-house followed by let's-just-buy-something-that-works. MB seems to be in the in-house part of the cycle.


Maybe because we as customers don't really need a complete corporate consolidation and control of every product by a few software corporations?


So they're 2-4 years away from a big announcement that they're becoming more efficient and maximizing shareholder value by not developing their own OS any more and instead they're "collaborating" with Google and shipping Android Auto. I bet chatgpt could write that future press release right now and get pretty close.


Everyones talking about alot of negatives. But I have a E53 AMG 4matic+ coup. It looks amazing, like constant compliments, i've even had people compliment it and give me their number in a wendy's drive through.

But the OS(MBUX 2) is actually really good, at first I was a bit negative on it. But the way they have everything integrated that isn't really possible with apple play is nice. The dash is customizable, I have my setup with speed, an array right in the center on when to turn, and the right is a mini map so I can see the road. On top of that the screen in the center is amazing and has a few views.

The voice assistant is really nice, does everything I want, they have it baked in so each seat can activate it and it knows based off a mic. Can't do that with apple play. So when a girl in my car says "Turn on the massage" or "Turn the temperature to 70", it only does for their side. And honestly, besides the map not having police checkpoints, it's a bit better than apple maps or google maps. And on top of that, I can lookup the location on apple maps on my phone, then send it to the car.

And according to my dealer, most people forgo using carplay for anything other than music in alot of the benzes.


I'm blown away you think "[women] give me their number" is like... something we ought to know about using the home grown Merc infotainment system? Like that's a feature I look for in my vehicle onboard computer, a "propensity for women to want to date me while I order fast food" metric.

So many questions. So many.


The infotainment and lighting + other controls really sells it…

I more meant that even from a style standpoint, I never had that happen when I drove a Honda. There’s also features like making the engine “purr” which works really well on dates. There’s just a lot of integrated features they can control, to the point that not only is it useable and I like it better than CarPlay. It also has tangible benefits and genuinely helps me score. With features that wouldn’t be possible with a CarPlay based os and not having everything super integrated.


> There’s also features like making the engine “purr” which works really well on dates.

100% it just plays some sound via speakers. Nothing to do with an engine. But a sale has been made.


It has a few things like this. https://youtu.be/mG3Uo0a8k5Q . It basically opens up the valves/exhaust a bit more for more noise. It's 100 % from the engine and not from speakers.


Okay. Good, so it’s mechanical. I was ready to sacrifice my arm so good to learn about this beforehand :)


If it works consistently, it would sure be a major selling point. ;)


Wouldn't someone giving you their number based on the car you drive be a giant red flag?


People who drive those cars don't think it is.


People driving a "coup" are also a red flag. So I guess it's red-flagged people meeting one another.


Not with a Mercedes.


Right? Like Alan Harper said: never happened with a Volvo!


I have an EQB and I totally agree. I insisted on getting CarPlay but I used it once and I don’t see any need for it. Much happier with MBUX. However, this new Android based OS isn’t good news and I’m not sure I would want to get a new car with an Android based OS in it.


It's likely using Jolla's AppSupport (no official announcements, just my guess). It's not Android Based but an Android compatibility layer.

Kind of how WINE lets you run Windows apps in Linux. I have a Sailfish OS, and it's a linux system (full command line, UI built with QT) but also has Android support via their translation layer.


Android Automotive has been getting very good reviews in vehicles that has it (mostly because its performance is pretty good).

If you look at e.g. Renault, it's also highly customizable by manufacturers so it's unlikely you'd even notice it's Android without looking into it.

Although, as far as I know MB isn't on board for it - BMW is though.


I rented an Audi and this was my experience too. It ended up working out really well to use the car navigation system and I still had car play right there for carplay and phone calls. And for navigation stuff you can still load up google or apple maps if you need to double check the route since they sometimes have different traffic information. But most of the time you don't need to


I was in the same boat, purchased my car with CarPlay because I thought I'd use it just as often as I do in my last car. MBUX is honestly good enough that I don't bother plugging in my phone, it does everything I want, integrates with the car, and stays out my way.


Are you showing off here?


No, I'm just trying to build a point that the high end car manufacturers are able to get some wild integrations going on that differentiate them from regular manufacturers due to their software. Especially Mercedes, who regularly has the best tech in their vehicles, years ahead of the other manufacturer, and usually a year before BMW or other high end ones. This makes a very nice product. Since otherwise you run into an issue of "Why would I buy this", when the reason people get these cars is how integrated they are and the attention to details.


So a big flashy glowing dick extension. It strokes the owner’s ego.


Sir, this is a Wendy’s.


>Offering flexible upgrades

do you want to buy the headlights DLC ($0.9 first month, then $29.9)? Y/n


Please press the brake or accelerator to confirm your subscription


Or just wait until ADAS front assist gets a false positive around a t junction and someone stops in your rear.


I mean … following too closely gets you liability for the collision and typically a citation.


With exceptions, brake checks can put the car in the front at fault.

Regardless, it doesn’t help when the car is damaged and there was no reason for aeb to kick in. The computer thought it was right but it wasn’t.


Sorry, your package only allows you to turn the volume up to 45%, would you like to upgrade?


I'd personally love if manufacturers actually offered upgrades on their older cars. I had an older Mercedes-Benz and I would've absolutely given Mercedes-Benz money if they had a first-party CarPlay solution for my then car.


Fun fact: Porsche offers this for some of its older generation models.

https://newsroom.porsche.com/en_US/2023/products/porsche-cla...


Yeah, Porsche is the only manufacturer that I know of that actually cares about older cars they've sold.

They even made an amusing ad to that effect: https://www.youtube.com/watch?v=R8-9oIq1hxw


Mazda upgraded 2017-2018 systems with carplay after sale. It cost some money at the dealership but it was a standard procedure they could do


You're effectively saying that you want more disabled features, just in case you want to upgrade your property at some point.


No, they're saying they would pay for new software. Many ~10 year old cars have all the hardware necessary to support things like CarPlay, but because CarPlay didn't exist yet they do not have it. A lot of people would probably happily pay $200 to add it to their car without having to shoehorn in a third party head unit.


I think it was more about offering features on cars manufactured prior the new feature was released...


Not really. CarPlay didn't exist when the car was made. The car manufacturer had to add software integration even if the hardware was already competent. Parent comment is saying they'd pay to receive this added integration.


Pop-up ads over your fuel gauge. Monetized!


Turn left over there and stop at gas station!*

*- Proudly brought to you by Shell.


DRM-Lock gasoline! , you can only use MB approved partners :), our nearest partner gas stations is just 5 hours away.


I can't wait when infotainment will start playing ads.


If you use Waze for navigation, it already does play ads when you are at a stop.


Mercedes has been disappointing me over and over again. Their car body designs are becoming difficult to look at, especially the electric range (the EQS is one ugly car). Their software is HORRENDOUSLY buggy. Their 2007 S-Class is a smoother ride than their 2022 model.

I love what Mercedes used to stand for, but right now they are pretty much selling cars on brand value, not because they're actually great cars.

The new S-class even feels like a step back from the previous model in terms of basic aesthetics. Plus, loaded up with cheap electronics, ready to break before your next service.


Those cheap electronics are the main selling point of the car. Flashy, blinky, lots of nvidia chips. That’s all there is to it.

It happens to all brands slowly. Cars have no longer any excitement to offer so manufacturers load them up with this cheap crap.


Comments like this actually turn me off buying MB again. I used to have a lot of MB cars back in 90s- 200xs - C-klasse, S-klasse, and then a number of V-klasse and every time it was a pleasure in every sense. Never had any unreasonable problem with them. Buggy software? It was unheard.


Then again, this comment also calls EQS ugly (which it aint) so maybe take it with a grain of salt? :P


I think you're the first person I know who likes the look of that car.

Edit: For over 100k GBP, there are other significantly better options.


I do like it as well. It has MB character, whether you like it or not.

What are those significantly better options?


Good point :)

Those things are quite subjective, I am just trying to avoid fruitless discussions


> The company is confident that this strategic approach to software and hardware development will be the basis for lifetime revenues as well as additional contributions. Already in 2022, Mercedes-Benz generated more than 1 billion euro in software-enabled revenues with products and services such as navigation, Live Traffic or online map updates.

Get ready to pay a monthly subscription to use features that come stock on your car.


What's the legality of hacking these things, assuming you own the car?

For example, BMW charges $18/mo to turn on heated seats you already own. Am I legally allowed to hack their software to enable heated seats w/o paying? How about installing a wire and switch and running them manually?


Well, depends on how much do you want a 5000-15.000$ warranty repair on your 80.000$+ vehicle to be rejected because you messed around with it.

It's not about difficulty of hacking, but with just how easy it is to put in DRM measures that will make sure that official repair shops will reject the repairs (and unofficial ones are really struggling due to lack of access to DRMed service tools necessary for new modules).


Of course you are, otherwise jailbreaking would be a crime.


All it needs to do is support CarPlay. Honestly car manufacturers should just let Apple do the whole car takeover. I’ve not met a stock system I liked. I’d rather it be powered by my phone.


I added a $200 CarPlay radio to our old Honda and it blows away every factory nav unit for usability (and sounds adequate). At this point, I’m only buying a car that has, or can easily have added, wireless CarPlay.


I like the BMW and the Tesla ones. I also heard good things about the Polestar Android.

The current OS on the Mercedes looks very laggy and not very user friendly, but if they fixed these major issues it could be quite nice. The high end Mercedes has a huge oled dashboard, it’s not something Apple CarPlay supports.


Tesla's is fairly good, but the way they handle text messages leaves me really wishing I could just have CarPlay. I may install a third-party screen somewhere so I can have it back.


Oh yes, that's what I want - permanently online, bug ridden OS playing me ads and forcing me to buy software "upgrades" of stuff already installed in my car.

What a Brave New World.


The use of touchscreens has made handling devices and machines a lot less pleasant. We will soon see a reverse movement, where physical buttons become more important again. So much more satisfying.

Of course, this can still be combined with plenty of software.


I'm surprised the StreamDeck family of gizmos hasn't caught on as a popular aftermarket car mod. It's just little LEDs inside clear buttons. I'd love about 5 of them right on the center console that I could somehow program to do routines for me.


Those decks are a single screen with a mask and mechanical button overlay - you can argue that there is additional cost to a regular screen without buttons, so maybe it's additional cost they avoid.


Yikes, I don't feel like Benz has the Software Engineering talent to pull something like this off, which will definitely make me hesitate in getting in a car that runs the OS


A lot of people will (wisely) wait to see how this pans out before getting a car with this software in it.

The car industry is full of manufacturers who decided to build their own software and failed miserably -- this continues to this day. With the exception of Tesla, very little car software matches anything like CarPlay or Android Auto and that's basically a medium sized feature of your phone.


>matches anything like CarPlay or Android Auto

Those are not any OS, just a basic graphic overlay. OS for instance would be the one communicating to AndroidAuto over bluetooth/usb, and driving the GPU that renders it.


This just seems doomed to failure. It’s not fully open. It’s not fully closed. Trying to combine too many things, and the google tie-ins are a turn off for me.

Automakers don’t make good software or UIs. But they refuse to yield that to truly competent suppliers.


“In China, Mercedes-Benz is engaging in a local cloud partnership with Tencent to support its automated driving systems. An enriched UI for automated driving functions including an advanced lane-level map view is planned.”

Tencent, really? How did that come about?


This is how deals are done in China. You want to expose your cloud service to Chinese users? You need to partner with a Chinese company that will effectively run the cloud infrastructure, possibly run the server software, and definitely have access to all of your IP and certificates.

The real question for me is what does an ADAS system rely on the cloud for? Hopefully nothing in real time.


The concern among many eastern countries regarding cloud usage is location of data storage. ADAS system could rely on cloud for - Getting updated maps (HD in some cases) - Storing/uploading camera data for analytics/replay among other things...


> This is how deals are done in China

You seriously think deals are done differently in other countries if you are as big as Mercedes Benz?


Yes actually, I have experience running a cloud service where we had to deploy two environments: China and ROW (rest of world). AutoOEM of the same magnitude as Mercedes-Benz.


Ok... What car brand has the least bad UI these days?

I have been putting off getting a new car for about a year.


I have a recent Mazda.

It boots to CarPlay pretty quickly. Its native controls are pretty utilitarian when they are needed (e.g. to switch to native SiriusXM when cell signal is spotty). I even prefer the look of its maps although I don't use them. Nothing in the car is a touchscreen.

And, get a load of this: they still have actual and intuitive physical buttons for climate controls, defrosters, and driving modes.

Oh, and the car itself is decent as well if you forgive the meh fuel-economy.


Not only that, mazda has properly working buttons for infotainment: if you press "nav" or "music" it will bring you back to last one you used. So if I use FM radio and Android Auto maps - buttons leads me where I want them to lead me. So many cars had physical buttons, always lead to built-in apps instead of Android Auto and Car Play.


Do they have analog dials for speed and RPM?


Mazda. They’re all in on being anti-touchscreen plus they support CarPlay (with a physical control).


One with Apple CarPlay and/or Android Auto built-in, so you can avoid the car UI as much as possible.


I put a quad-lock on my dashboard and just stick my phone on it. It's right next to my steering wheel. I set the stereo to bluetooth link and everything's actually great.

I think CarPlay and Auto are too much. Outside of navigation and music, what overlap do we have in a car and phone? Just enable bluetooth sending of locations or full-on directions. Bluetooth music already works great.

Eventually we'll have voice agent integration so that stuff nearby (e.g., bluetooth range) gets integrated into the available status/command set of one primary one.

Leave all the fast-moving, upgradable stuff on the phone. Make the car platform stable and you've got the best of both worlds.


Totally valid. What's a quad-lock? is that some sort of phone mount?


Yeah it's a special connector with phone cases and mounts for cars and motorcycles. Works great


In my experience driving plenty of rentals (EU-centric), Mazda, Mercedes, Audi have all been pretty good.

Volkswagen was a standout for absolutely awful experience across the board with unituitive layouts, horrible capacitive buttons and gimmicky lag galore.


Probably an Ibex [1] as they're based on the post-2007 Landrover Defender dashboard

[1] https://www.ibexvehicles.com


IMO, Hyundai and Kia.

They currently have a bad rep for lack of engine immobilizers in non-current vehicles, but they've nailed the UI with a combination of software and hardware buttons.


Tesla, I think


Maybe visually. Not practical


The MBUX is awful. Unresponsive, lagging, buggy, unstable, anything that could go wrong is going wrong. Looking forward for fixes or more interest towards the user.


These people should be banned from making software. Some imposed penance would be good too.

“MBUX” is buggy and confusing as heck in a brand new 2023 model.

Apparently one of the most requested features in new cars [1] has been CarPlay - people are tired of the native “infotainment” garbage

[1] It's also a top feature for many drivers and car buyers. Twenty-three percent of new car buyers in the U.S. say they "must have" CarPlay https://www.cnbc.com/amp/2021/05/29/apple-carplay-massive-su...


MB.OS stands for Monetization Backdoor Operating System.


Is this using Jolla's Android AppSupport? https://jolla.com/

I use a Sailfish OS phone and it works amazingly well. I know they have been pushing hard into the auto industry after pulling out of Russia because of the war. I would be very pleased if Mercedes is helping fund Jolla.


In the past few years Jolla worked together with MB. In the latest Sailfish release there were a lot of Android updates, and it seems of high importance to Jolla. At the same time, Jolla could use a new investor, but there was no news that MB would be interested in becoming an investor in Jolla.


I'll give Google their due, their satnav maps apps was accurate enough and fast enough to not need a co driver on unfamiliar roads with my Colin McRae head on. Thats more than can be said with MB's Becker Maps.


It might actually be nice and useful if automakers operated portals that received telemetry from all their vehicles and offered owners statistics on error codes etc. It would improve the automaker processes too since the supply chains for parts could be need based, the part waiting at the dealer for the malfunctioning car to arrive rather than sending parts in response to orders.

I'm aware that there is a lot of economic alignment to facilitate this but I'd like to think the future of car software could actually be helpful to people rather than renting heated seating.


> Building a proprietary system: The Mercedes-Benz Operating System (MB.OS) is a new purpose-built chip-to-cloud architecture that will be central to every future product, delivering exceptional software capabilities and delighting customers with its ease of use

Say bye bye to actually owning your vehicle or having any kind of privacy. Its really time for car jailbreaking.


>Unlocking financial opportunity Lmao, MBAs. If you don't want to spend so much money on writing software then FORK ANDROID. Or, you know, make a super dumb entertainment unit that requires a phone, AA/CP support and then you're done!

Can't wait to see all of the super dangerous vulns that come out of this over the next wee while.


I just found this presentation with technical details: https://group.mercedes-benz.com/dokumente/investoren/praesen...

I guess it is not a typical OS, but more like a collection of tools that build on already existing OSs. This is comparable to ROS (Robotic Operating System), which are just some programs, middleware, services and conventions to build software for robots on Linux.

It seems like this should integrate and abstract different "OS" (Linux, QNX, AUTOSAR) and run on very different platforms (high power application processor for infotaiment, microcontrollers) (Slide 13). These are widely different systems:

1. Linux needs a Memory Managing Unit (MMU), which only comes with high(ish) powered application processors, e.g. Arm Cortex-A9. These are obviously not hard-realtime because a page fault can occur non-deterministically (except when you can lock everything to RAM). This might be used for infotaiment.

2. (Classic) AUTOSAR is used without a operating system on a microcontroller like the ARM Cortex-M or a automotive MCU like the Infinion TriCore, which can run two cores in lockstep to verify each computation. AUTOSAR is kind of the operating system and you buy an "adaption"/HAL of AUTOSAR to the each MCU from a vendor. This is widely used in many ECUs for hard real-time control, e.g. to control something in the engine, and other stuff like the electric windows. AUTOSAR is a huge pain in the ass to develop for. You usually configure "it", which takes a lot of time. Then a software generates a huge amount of code. The software is from another vendor, e.g. Vector or Elektrobit. The developer fills out the function stubs implementing the actual function. Alternatively, you can generate the code from MATLAB/SIMULINK models with a code generator from yet another vendor (model-based-development). The upside of this, that the HAL and code generators are certified and everything is somehow standardized. The downside is that normal developers want to kill themself, you learn no transferable skills, and the huge amount of generated boilerplate code, that is hard to read.

3. There is also a newer Adaptive AUTOSAR, which can run on Linux or QNX.

I guess (page 8) they want to use it for infotaiment (point 1), interior control (lights, climate control; probably point 2), automated driving and "central driving" (point 1, point 2). I am not sure if this includes typical fast hard-real-time tasks like engine control or chassis control (=vehicle dynamics control).

I am not sure if really want to abstract it all or just extend the "OS" (Linux, QNX, AUTOSAR) with libraries and components, mostly in user space.

If you look at slide 13, you might guess that they will adapt Linux and QNX to run their UI MBUX (in QT). They extend it with services that communicate with ECUs in the car and their servies in the internet. Moreover, they allow to install sandboxed apps from Mercedes, Android Auto (e.g. Spotify) on top of it. They also come with an app store: https://faurecia-aptoide.com/

The real-time ECUs in the car running AUTOSAR will just get additional components to easily communicate with other MB.OS parts and support some newer features like OTA update.

I have not seen any details how this relates to ADAS functions. These are typically (partially) run on a compute node made by an automotive supplier with a hardware accelerator from NVIDIA, e.g. ZF ProAI (https://www.zf.com/products/en/cars/products_64166.html) or from Valeo (https://www.valeo.com/en/domain-controller/).


Just a nitpic, Linux can run on CPUs without MMUs, and it has APIs for locking memory and real time scheduling but that's not why it is not a hard-realtime OS.


Yes you are correct. I am currently having a project, making Linux "as real-time as possible": locking memory with mlockall, isolating cores, preempt kernel patch, .... It is still not real-time because you have no guarantees, but you typically get a max jitter of 0.1 ms, which is good enough for my use case.

You could use Linux without a MMU (uClinux), e.g. on a Cortex-M, but is a horribly experience and no standard program works.


Preempt RT does give guarantees in the sense that unbounded latency is a bug and theoretical maximum latency bounds are known (see [0]). It is neither certifiable nor formally proven, but it's good enough for almost anything that isn't safety-critical. For the things that do require functional safety, you can use AGL and other hypervisor architectures that partition the critical and non-critical tasks with a few more changes to your code.

[0] https://bristot.me/demystifying-the-real-time-linux-latency/


Preempt RT doesn't really give that. It might give that when you run a subset of Linux, but that is not Linux like nommu Linux is not Linux. They might say that's a bug, but there are countless algorithms and data structures in Linux that mean the state of the system and other workloads can slow down other parts of the system. Even setting aside the fact that a lower privileged process can take spin locks (not "preempt spinlocks" but real low level spin locks), disable interrupts, etc., they can influence shared data structures such that allocations, lookups, etc can take longer for the higher privileged thread. So you still end up with a "look we tried really hard and if you don't use any kernel facilities including blocking and isolate this CPU entirely, lock everything and don't allocate memory or take page faults after that, you might get something approaching hard-realtime".

People try to paint it as soft-hard-RT or something, but it's not, there already exists a good word for it which is soft-RT. Which is fine, it's highly useful.

There aren't really formally proven hard realtime operating systems of any non-trivial complexity are there? They are either extremely simple executive layers, or some very limited privileged functionality that sits on top of the rest of the kernel.


I'd highly recommend you read the link and the paper it's based on. It's pretty thorough in addressing the limitations. Those same limitations apply to virtually every commercial RTOS out there as well though.

As for formally verified systems, depends on your definition of "nontrivial". You can build complex systems from the building blocks provided by the well known examples like SeL4 and pikeos. On a practical level though, complete formal verification is incredibly uncommon for exactly the reasons you'd expect. There's usually a mix of formal methods and other verification methods employed in safety critical applications. It's "good enough" given current capabilities.


I did read it. I understand and work on Linux including real time Linux. Nothing of what I said is wrong. Hard realtime operating systems of course are more limited than general purpose Linux too, but they tend to have a much better handle on limiting and controlling latency and how non-critical workload can impact critical tasks.

And seL4 is formally verified but as far as I know it has not been formally verified for hard realtime. Funny thing about formal verification is that it's easy to do if you control the requirements :) (/s - nothing to take away from the incredible work of sel4). Last I heard people had sketched or theorized about ways it could be approached, but not done.


Very interesting. Thanks for the link to the paper. Isn't the provided paper "just" about the scheduler? Eventually, I would have to output some data, e.g. on the CAN bus with socketcan through the network stack. This is probably a huge amount of code for which worst-case-execution-times are probably hard to get.

Does AGL mean automotive grade linux? What would be other hypervisors?


The time it takes to put things on a physical bus will depend on your hardware and can be bounded, but this isn't the guarantee you're getting from any RTOS.

The main thing the "RT" in RTOS guarantees is that the OS will return control back to you in a defined amount of time as soon as you're ready to run. You're still responsible for ensuring all of the other system requirements for bounded latency are fulfilled, like hardware that doesn't introduce unbounded latencies the OS can't control (surprisingly difficult with modern HW). Assuming you've done all of that, preempt-rt will give you essentially the same guarantees because of the scheduler work linked.

Yes, AGL = automotive grade Linux.


It's notable that Elektrobit seems to work on getting Rust on QNX:

https://github.com/rust-lang/rust/pull/106673


Hard to believe a company would hawk a product in 2023 by first proudly pointing out that the thing is proprietary.


Now the shake out of car companies that make software vs software companies that make cars. Reminds me of the old bellheads vs netheads. [1]

[1] https://www.wired.com/1996/10/atm-3/


"Building a proprietary system:..."

This will pretty much reassure everyone willing to drive under the control of such.


> The next generation of SAE Level 2 automated driving will start in the Entry segment, leveraging the full potential of machine learning.

Forgive me, but my expectations for the "full potential" of machine learning are not so high. They're definitely not high enough to trust my life with.


It doesn’t matter. The legislator is sold on this, it’s required by law, nobody cared what we think. Number of accidents goes down annually and nobody cares that edge cases like you or me are pissed off that we’re no longer in control. Because, you see, apparently driving a car is a privilege and not a right.


Mercedes-Benz UI for cars circa 2015 has to be the worst piece of UI I have used in the last 30 years. It is slow, unintuitive, has both touch controls and a turning knob that has no semblance of sanity, and each has exclusive UI control properties.


> The company is confident that this strategic approach to software and hardware development will be the basis for lifetime revenues as well as additional contributions.

Why sell cars when you can sell SaaS? Get ready to pay a $49.99/mo. subscription for access to power steering.


Is Mercedes actually building an entire OS, or does this really have something off the shelf underneath?


All major car OEM are challenged with a cohesive software development and integration. There are so many different units of ECU/ICI/ADAS/TCU/ABS/etc (and list goes on) and mostly all of them are interconnected, partial OTA, etc. Tesla having started software first and with their central ECU approach has a leading edge.

Mercedes with their MB.os, VW with their Cariad, Toyota with their Arene and each major automotive OWM is coming up with their own overarching OS platform shifting to software and mobility first.

Underneath is often a safety certifiable OS like QNX, combined with AutoSAR or Linux for less safety critical OS. So generally as far as I can tell most car OEM do not start from scratch and they have to work with the leading silicon supplier as well on all the OS driver support. You also have to consider full functional safety compliance across the globe, many teams/languages/etc. It truly is not that easy to orchestrate that all.

On top like someone commented here, many OEM are very hierarchical structured and creating software development across these different functional units/devisions is a challenge for most automotive companies are really struggling with.


Is this a new kernel, e.g. realtime? Or BSD/Linux with a new graphical shell on top?


I don’t know if this was written by a computer or a person, but the result is the worst kind of corporate-speak.

“MB.OS is designed to connect the major aspects of the company's value chain.”

I gave up trying to extract any meaning from this article pretty quickly.


'Unlocking financial opportunity: Software-enabled revenues to evolve to high single-digit billion EUR figure by end of decade;'

Does this mean MB owners will have to pay subscription fees to unlock, eg., heated seats, in their vehicles?


all sounds good…to be honest read some paragraphs but not whole page.

step 1 for Mercedes Benz is to make a vehicle reliable especially at their asking price.

had a mb c300. lots of oil leak after warranty period. sensors also bad, constantly complain about low coolant level, even when it good. transmission also slipping.

purchase car brand new. never abused engine..always put high octane gas..did all routine maintenance from mb dealer.

Will NEVER buy mb product. when i read about new software to me they just bunch of fancy talking clowns. have a 2012 hyundai elantra now driven by me daughter.

more reliable than mb c300. even hyundai more reliable than mercedes benz.


I wonder why would they undertake such an Herculean task when they could simply pick a non-GPL OS, or just pick Linux itself given the programs running in it wouldn't have to be open-sourced anyway.


Ah, this will be fun to watch. I give this about a 3% chance of producing something the end users will actually enjoy using.

Hardware manufacturers are generally extremely bad at building software teams. It's against their DNA. Their executive leadership culture tends to actively reject people who are good at leading software teams, nevermind how much they want that sweet UI/software. (Yes, Apple exists and they are a hardware company that's pretty good at software.)

Then add a German extremely hierarchical largeco culture to the mix...


It's strange seeing the word "proprietary" used as if it's a positive thing for consumers.


I can see it now, 2dollar monthly subscription to the glovebox compartment


> Unlocking financial opportunity

bazinga


> …and receive offers tailored to their interests…

The day my car starts showing me advertisements of any kind is the day I take a sledgehammer to its infotainment system. For a luxury car brand to even include this in their copy they send out is a bit worrying.


do you remove all their advertising badges, model numbers, logos from the outside of your car too?


Funny enough, pretty much all German brands give you that as a selectable option for like 100eur :D

You get a car with most badging removed.


Would gladly pay 100eur for zero badging.


Out of the park non sequitur.


What? Someone hates advertising and especially with their car. Absolutely reasonable position to take. I ask them how far they go with that? Reasonable to go absolutely hardcore and get rid of /all/ of it. Reasonable to say that level of advertising isn't that annoying so not worth the effort. I ask which because I'm interested in the response. None of that looks particularly controversial to me and I'm not seeing a non-sequitur.

I refer you to the guidelines recommended by the house which are here: https://news.ycombinator.com/newsguidelines.html

See what you make of them.


Do you not see a difference from a Mercedes logo on the hood and and McDonald’s ads blaring from the infotainment system?


Obviously they are different. You think they are beyond what I just painstakingly pointed out as one of degree and a degree that you are entitled to feel differently about each case?

How do you think your comment advances the discussion, just out of interest?

Fwiw I don't tune my radio to commercial FM. I do have badges on the outside of my car and I kinda wish I didn't but not enough (yet) to have done something about it. I don't wear clothes that turn me into a flipping nike/ralph lauren/chanel/gucci/whatever billboard either but I can see this puts me in some kind of strange minority but I still find wearing advertising without being paid really weird. Sadly badges on cars are a little harder to not buy or remove.

You? Where are you drawing your line?


The difference between those two things is kind of like the difference between skipping beef on Mondays and refusing to eat food that was cooked in the same kitchen as meat. It doesn’t really make sense to ask without context, right?


Someone says they are ethically opposed to eating meat. Ask them how far they go with that.

There ya go, fixed the analogy for you. Do you understand now?


Mercedes is high status. McDonalds is low status. To be able to see the Mercedes Star is a privilege.


Wearing mcdonalds logos on your clothes or car shows you're "important," "successful," and "influential" enough to be paid to do that.

Trumpeting a mercedes logo on your car to try and claim status shows you're something else.

Just a different way of looking at status, and one no less valid than yours. But I think we've digressed beyond it being a useful direction to continue.


Why oh why would they want to re-invent the wheel one more time?


EMAAS - electricity and mobility as a service

This is the new direction for car companies, they're getting out of the car business and into software.

What could possibly go wrong?

ERROR: vehicle unavailable due to network instability


ironically there aren't any screenshots in the post to show what MB.OS looks like.


It likely doesn't look like anything other than bitstream. Think 'linux kernel' rather than 'ubuntu'.

It's a foundation component. MBUX is a layer on top of MB.OS, and that does have a particular look and feel.


Lots of negative comments in this thread, which I can understand.

But let’s be fair, for a car manufacturer, carplay is the kiss of death.


How so? I mean, they're still manufacturing the actual car itself.


For software-eating-the-world reasons, I'd guess. A chance to differentiate your product as software does more and more, versus a big chunk of the experience being within someone else's control and the underlying vehicle being an interchangeable commodity.

I haven't tried it personally, but people seem pretty passionate about the software on the leading electric vehicle brand, plus its associated do-everything phone app letting you control the vehicle, find a charger, or even order a new model.

The situation will probably be like smart TVs for quite a while: lots of janky experiences and user-hostile behaviour, because it's just so valuable to own a software platform that every company tries.


Android Automotive is now the native OS for new Volvo and Polestar vehicles. A far better option than anything custom from the manufacturer.


Weirdly enough I never had issues with my previous generation Volvo that used their proprietary OS (circa 2020) compared with the Android Auto built-in my latest Volvo (2023). I've noticed that my backup camera app can be slow to load almost sometimes failing to load or even loading and then crashing when you shift into reverse gear. I don't know if this is Android to be blamed or some flaky code by Volvo but makes me nervous nonetheless. On the plus side when everything works smoothly, its great to be able to load a wider library of Android auto apps compared with the measly selection of apps Volvo provided previously.


Don’t get me wrong. Android and Carplay are heads and shoulders above the manufacturers’ software. But that’s really bad for car manufacturers because the car will become the commodity while the service they provide won’t.


You could already replace the stereo in a car if you didn't like it, but you still bought the car.


Not really anymore. Most cars don’t have a discrete radio head unit anymore. The radio is way too integrated with vehicle controls.


Maybe so, mine's 10 years old so I haven't checked more recently.

I assume you can stick a CarPlay supporting screen on top of the existing dashboard at least, but it'd be hard to integrate.


Let me guess, Android the hood


Ooh, Nvidia Drive platform.


is it microkernel written in rust?


Software-enabled revenues? Proprietary? Built in Google spyware? Holy shit sign me up!


Software is what legacy car manufacturers have been lacking so it’s a move to the right direction indeed. Curious also about the future OSes of BMW and VW Group.


I take any “legacy manufacturer” vehicle over the “modern manufacturer” car. If my only choice is VAG, I take Skoda.


Car is an appliance, like a fridge. I don't want to have an internet connected display showing me ads and "upgrades" on a fridge, nor a car, nor a TV.


VW tried to do their own software with the ID series and failed horribly to the point the CEO had to be replaced.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: