Your car is about to go Open Source | Hacker News Reader

Your car is about to go Open Source

computerworld.com

40 pointsjebblue31 comments

The customers don't want complex "informatics systems" in their car because:

- they have to drive the car, most of the time they just want ome music

- if they are a passenger, they can just use their phones

- the automakers sell them at outrageous prices - many times the price of a smartphone for a device that has less functionality

- the integration with other devices is difficult. The new Mazda6 has great Bluetooth integration with an iPad, but you don't get any of the features when connecting an Android phone.

Also, why that title? My _current_ car isn't going open source. My future car, _maybe_ will have _some_ open source software. It will not "go open source", just "use a small part of open source software". Also, "going open source" doesn't mean that everybody will use the same software and hardware. We will have fragmentation, just like Android has right now. Sorry, you can't have application Y because your car is 2 years old. Buy a new one!

I can actually comment on this which is rare. Now this isn't directly from the horses mouth I do work with the auto-industry but not directly in 'info-tainment systems'.

They are not going 'Open Source' any time soon, a few automakers may. And most developers would prefer it honestly. But currently Ford's System is build on Top of WindowsCE running on a PowerPC chip (which is basically custom for Ford I believe (thus the microsoft sync), I don't work in CE a lot so I don't know a ton about it).

Originally the 'Sync' system was going to not only handle info-tainment oriented stuff but car diagnostic information such as Valve timing and engine firing. The problem was the scope of the project was to large, and caused way to much CPU load, I believe its roughly a 600Mhz PPC chip, very under powered for the task at hand, not to mention windowsCE is a <sarcasm> LOVELY </sarcasm> development environment to work in (I know this first hand, different dependencies for a debug-able, and non-debug-able binary just make my heart sing with joy).

The problem they faced was that info-tainment system errors would cause it to short restart, or hang... just general stupidity due to first gen software problems. But when your also running your knock sensor on a different thread of your 'info-tainment' CPU, things get 'real' very very fast. How bad? Well when the computer when buggy the motor would default into 'prevent knock mode' and suddenly every single cylinder started firing 5+ degrees into the down stroke to prevent engine knock even if timing was correct3-4 degree's prior was proper timing. So they moved engine control away to a different tried and true processor.

Now whats left is this really buggy thing that limps along H/A'ing media, and giving you some car starts since its still technically on the CAN bus (which industrial busses are completely open source, you can thank people from the 1970's-80's for that, not out of 'creating a community' but so you can figure out what's being sent with a 19200baud scope, and an ASCII look up chart). So its 'buggy' but the WinCE on PPC was partially sponsored by Microsoft, who gets free advertising for keeping WindowsCE alive for Ford.

There are rumblings in GM to move to a Linux plate form because the developers are practically on strike over stuff like this, but GM moves very very slowly, and is more bureaucratic then the US government a times, so don't expect this to become a 'thing' in a time-span less then 10-15 years, because that's about how fast 'new' things get integrated into cars around the Big 3.

I'm rooting for you guys on the second floor.

Also not in this area, but I get the sense that there are a few mindsets to it:

Future is smartphones and the like, let's integrate well with that (ie Kia/Hyundai)

Future is our own thing, but can accommodate some other tech on our systems using back end servers as a shim(ie VAG)

Future is what we chose, will accommodate some bluetooth and USB though (ie Ford)

Future is let us try to get into many OEMs and to make it more appealing offer a developer program (ie OnStar)

Touch-screens in cars still feels like an impressively bad idea.

totally agree, especially when they're not tilted towards the driver and badly polarized to interfere with sunglasses.

drove my buddy's 2013 Subaru WRX, the "Limited" edition that has this touchscreen as a standard feature (as in, it cannot be opted out of) - a truly terrible experience. Just seeking radios stations is an exercise in anger management. Since you cannot feel where anything is, i found myself much more distracted from driving with hitting the correct place on the screen. Having a large screen adds to the frustration since you cannot even have good reference points while you hunt for what you're looking for. Any visual buttons that are not on the very edges of the display are just worthless to use (and there are lots of them). There's no muscle memory, every time it's like driving a different car.

I assume that if/when Glass takes off, Google will follow-up with an integrated car-version. Glass is meant to provide non-distracting visual info at a glance and operated hands-free, so it's an obvious platform for cars.

edit: I'm always surprised how little the consumer electronics industry learns from the gaming industry. Games have refined the modern gamepad into something impressive - they force developers to shoehorn their games into the given controls, and adapt those controls to the correct context, and give the user good non-distracting HUD feedback.

Meanwhile, you look at a TV remote or car stereo controls and we're transported back to the bad mistakes of the '80s. Rip out half the controls and just give me some good text-to-speech to tell me the context of what menu I have currently selected.

...and so we'll get more of "standard feature (as in, it cannot be opted out of)" - smart glasses!

As another related point, let me remind you that we're having the idea of wearing smart glasses to have a convenient display in front of our eyes. In cars we already have car's glass placed permanently in front of our eyes.

I don't literally mean having a glass on your face while you drive. I mean integrating the technology that Google has built for Google Glass into your car - the new text-to-speech system and whatnot. Attach the physical display to the rear-view-mirror or project it onto the windshield or something.

I rent a lot of cars and I've seen some really bad touchscreens. But the Toyota Camry has a great one (as does the Prius) so it's possible for them not to suck. Hopefully all the car companies will get there soon :)

Off topic, I'm always extremely amused to see cars branded as "limited". "Ooh, this car has more limitations! Sign me up!"

pretty sure the implication behind 'limited' is that it is a limited production vehicle version and is thus more valuable because it has certain features that will not be provided in the standard.

I used to work for Panasonic Automotive. These things are universally awful because their tooling is still geared towards hardware-type systems. They can use linux all they want, but if it still takes two weeks to turn around issues brought up in testing, badly, then the result will be invariably awful.

The technology exists to create feelable/texturized buttons on a touchscreen. Even though the screen is flat your finger feels a small bump that would allow you to navigate the screen without looking at it. Alpine has used this on a few of their head units. Not sure why factory touchscreens aren't making use of it. Patents maybe?

Systems with excessively complex interfaces of any kind in cars are bad. I drove a Sync-equipped Ford last year. The infotainment system and several functions directly related to vehicle control were operated with buttons on the steering wheel, but several things that were single-function in most older cars and aftermarket sound systems required looking at a screen or memorizing a sequence of several button presses.

Particularly bothersome, the option to disable traction control was buried four levels deep in a menu system.

That was the second biggest reason I didn't pull the trigger on a Ford Fusion Energi a couple months ago (first one being Microsoft Software running critical car systems)

While Microsoft has a well-deserved reputation for making unstable and insecure consumer software, think about any software you've used made by companies whose primary business was not software and ask yourself if Ford would have likely done a better job in-house.

Touchscreens in particular are bad because you can't operate them 'blind' the way you can with tactile interfaces composed of knobs and dials.

I rent cars a lot when I have to travel, and I've grown to hate anything "touch" in a car. Several manufacturers have adopted touch-based controls, even outside of a touch screen. For example, I had a Toyota Avalon rental a couple of months ago, and I was constantly, accidentally pressing buttons because I brushed them with my hand. When a car is in motion, it's hard to steady your hand. With touch controls, I find that I have to plant a finger somewhere as an "anchor", then engage the touch inputs with another digit. What a horrible mode of interaction.

I agree, especially when the touchscreen replaces physical controls. I really like that in my current car I can adjust fan speed, temperature, stereo volume, etc all without needing to take my eyes off the road, but if I bought a new car with those controls moved to the touchscreen that would no longer be possible.

Also, most touchscreens are not that well suited for use in cars. The interface is slow is usually slow, clunky, and modal, the screen is hard to read in certain light and/or blinding at night, animations are often distracting, the touch sensor is resistive and ignores input if you don't push firmly enough/hold your finger long enough, etc.

It's disappointing that touchscreen interfaces are aggressively being pushed downmarket, to the point where it's very difficult to find a decent car without one nowadays.

I'd much prefer an open-source ECU to an IVI.

Agree. I was seriously frustrated recently while trying to get a friend's car to tell me why the brake warning light was on. Turns out a proprietary scan tool is necessary. A used one on Ebay cost $500.

Modern computer-controlled cars are incomprehensible to many traditional mechanics, but I bet the average HN member wouldn't have too much trouble understanding the code that runs the computer-controlled mechanical systems of a car.

Exactly. ECUs are notoriously closed-source and communicate using opaque and proprietary protocols, especially since vast swaths of them aren't even managed by the auto manufacturers (Honda et al just buy ECUs, do some cursory functionality verification, and off it goes).

Fuck infotainment in cars, at least until self-driving cars become ubiquitous. Nowadays it's common to have a dozen ECUs in any car made in the past three years.

The security in these ECUs is definitely a concern, especially since lots of them have 3G/LTE radios in them. And there are hundreds of thousands of these cars, everywhere across the planet. It's a massive attack surface that definitely needs attention well before self-driving cars get rolled out to consumers.

These new systems are trying to get people away from using their smartphones for navigation, music, social media, etc. The problem with cars today is they simply can not compete with what our smartphones offer. Google maps on my phone is far superior to anything found in a cars gps system and much easier to use while driving. A standard infotainment system that developers can program for can allow applications we are already using on our phones while driving and instead have it displayed with the intention of it being viewed while driving. High end vehicles are attempting to go autonomous soon (Mercedes is attempting by 2020) and users will want to stay connected since they won't have to pay nearly as much attention to the road as they do currently. Businessmen who commute to work every day can start working throughout their hour long commute.

I know what you're saying but, "The problem with cars today is they simply can not compete with what our smartphones offer.", I think when the car makers get away from locked, closed, proprietary operating systems; people might be more willing to put their confidence in the new software that will undoubtedly be developed.

As with most consumer prducts, the limitation isn't technical, it's what will drive sales and increase profit.

If you figure out a slick new way to mirror a phone's display, congrats, you've just cannibalized a major brand differentiator in the sales process. Better to require an updated car surrounding that updated software.

It made no sense to me a few months ago when I experienced this issue test driving an Audi. The salesman told me to hold off on purchasing a Q7 because it didn't have the updated software yet, but that next year's model would. Said the Q5 did have it now, so he was advising the purchase of a cheaper car due to the software stack.

Clearly there is room here to innovate. I believe Apple is facing this same issue with their upcoming Apple TV product.

It's almost better to make it even more barebone. You can use Google Maps / Waze / Apple, etc., for your navigation on your smartphone and it can also play music. You can even stream music.

I think it's much superior system than trying to make car work like smartphone. Adding connectibility in the car implicates that there's extra mobile contracts for cars, too...

I guess these days, those systems are also used for integrated components, like backing camera, etc., I feel it's more productive for automakers finding the way to make that work with phones...

What, why don't they just put an android tablet in a car console. Use a car launcher that focuses on music, navigation, and maybe read aloud texts.

Speakers, a microphone, possibly a charging point and perhaps access to a better aerial.

I don't really want a computer in my car anymore, i did in the 00's but today i just want very simple integration with my phone.

Touch-screens in cars feels like an impressively obvious idea.

Let me expand. From a manufacturing and design standpoint, it would be fairly easy to build some standardized components for touchscreens that different car companies can implement. They can then differentiate based on software (taking a leaf out of the mobile playbook).

They're already front-and-center (literally) in Tesla's Model S.

Exactly, and it's amazing. POC is there. I don't think there's anything terrifying about it.

Reply on news.ycombinator.com