Automotive Grade Linux - Already In An Infotainment System Near You
I was doing some of my catchup research during lunch today and decided to check on the progress of a system I had first learned about 2 or 3 years ago: Automotive Grade Linux.
They have been developing, per typical Linux fashion, an open source architecture/software for infotainment systems and possibly more. I admit I have only played with the system marginally as my programming skills are mediocre.
Other than the impressive list of members, and the fact that Toyota has already implemented it in the 2018 Toyota Camry, it would seem that software architecture wouldn't be a huge concern for us right?
Once one starts thinking if the current state of affairs though, software architecture does make a rather large difference. After all, wasn't FCA's hacked jeep accessed through the infotainment system? I don't currently recall what software was running that system, but I don't believe it was a Linux variant (android would be one variant). I am currently contemplating the implications of implementing Linux alongside SGW, and that is not a jab at Linux or its stability/security. I run a couple different distros on a couple different laptops for very specific purposes, I've always loved that little penquin.
Open source has been constantly touted as a way to prevent any one entity from controlling information and technology, the big discussion currently being around open source AI. I believe they have their pros and cons as anything does, but it is interesting due to the flexibility of the core architecture.
Added: Open Source Scope
It is also interesting because having open source code that the public can modify, patch, exploit, etc... seems to fly in the face of the thought that locking down software and access is the only way to maintain vehicle security (per my interpretation of current trends, always open to alternative takes).
My questions are:
- Does the software architecture of any system in the vehicle currently influence your job/diagnostic routine? Do you think it may in the future?
- Do you see this as a step forward towards more open source automotive modules and such? See George Hotz and his open source goal for semi autonomous vehicles.
- Do you believe that software debugging and/or modification will be a part of the job of future technician/diagnosticians?
Added Resources: For anyone with interest in such things, here are some additional resources available with support on the OEM level. It seems that the manufacturers are quite willing to open up at least some aspects of their vehicle software architecture.
Mercedes-Benz - Need a valid VIN number to access
Chris, I honour your open-ness, courage and contributions. To answer your questions: - YES. - No comment. - YES, It currently is in mine. Question: Could you please clarify your use of "distros" please? Thank you, Jaxon
Jaxon, Distros are distributions, various forms/builds off of the base linux kernel (architecture).
Hello Chris, I have not asked my question with enough lucidity: The couple of distros that you operate: which distros, and for what purpose are they utilized? Thank you, Jaxon
Jaxon, I run a laptop with Ubuntu 18.04LTS for general usage and playing with programming. I also run CAELinux (computer aided engineering), scientific linux, a second ubuntu based distro that essentially runs ROS (Robot operating system), andParrotOS. Main reason for multi versions is that each is precompiled and only needs minor tweaking. If I want to play with data modeling, I'll fire up
Chris, Thank you. Jaxon
I have no great skills in computing. Engineering fascinates me, as well as math, but they are means to an end. I utilize the programs to help teach myself concepts and then use programs like Parrot to find the flaws in basic programs I've written or to see how they work from a different vantage point. For example, I have been experimenting with basic neural nets as a means of predicting or
The one upside to mainly working on one line is a standardized diagnostic process, though there are small changes depending upon the year/system. While non-critical open source seems feasible, I don't really see things like open source autonomous going very far due to liability issues, or rather lack-of-liability issues. The "if no one is driving, who's fault is the accident?" problem is going
Chris, I agree with the liability thought though I'll stay out of legal territory on this, cousin Vinny will handle all legal questions. My only comment will be to think this way: home built from scratch vehicles and being able to register, as well as drive, them. My main thought, in this particular post, is that with increasing interconnectedness of all systems in a vehicle there seem to be
Hi Chris: Three quick points. When I was a kid, Keystone AAA used to send their insures a replica license plate key chain. I remember asking my grandfather why he threw it out. He said that if you lose your keys, your car was easy to find. Once your car was found, it was a no-brainer finding out where you lived. Then, not only do they have the keys to your car, they also had them for your
Guido, Perfectly phrased and exactly what I was hoping to have brought up. It is the same with the the encryption standards, be they AES 256 or RSA. It is not a matter of when they can be broken, but rather if it is in fact worth going through the effort to crack them for whatever questionable payout one may get, especially when pulling from a smaller group of users. As I had mentioned
The discussion (predictably) turned towards open-source and related security aspects, so I will try to take a look from a different angle. Linux is an operating system, which means it provides - an organized file system that can be used for extensive logging, - convenient updating, - standardized interfaces to peripherals (car modules), and - permissions management. Imagine this: in a
Great point Dmitriy. It is pretty neat to think about some of the fun one could have while using a linux platform, including but not limited to taking the error logs you mentioned and running a routine to translate that into a data set of your choice, be it strictly numerical or graphic. I believe you would greatly enjoy the data analysis potential that holds. My understanding is the majority
Oh Gawd, the GM30s are back. Ruuunnnn! If someone like Marty says these were good systems, drug test them! They have the good stuff. Think the plan was to balance the vehicle out? You had an engine that leaved like a sieve (though they got much better at that in later years). Maybe they thought that it was like a Harley from that era. It won't leak if it don't run. Let's add space age
Guido, Thanks for that link. Definitely some interesting topics being discussed, especially the symposium on AR/VR, though the ride to Toronto is a bit excessive. Maybe I can make my way blind baggage style to a few of them. I was not familiar with the GM30, but just started reading a 1987 paper referencing them and the introduction of the Tech 1. Even back then they were calling out the need
Hi Chris: You'll see an alphabet soup of O/Ss coming out of the woodwork when this is approved next year. (Registration required.) infoworld.com/article/335341… HTH, Guido
Dmitriy, I think OpenXC is right up your alley for data acquisition and analysis. I added additional resources at the bottom of the OP.
Good find, Chris! It's an attempt to go in the right direction, but it has limited advantages over polling with OBD interface without car modules actively pushing their data to the car computer to be stored (even before the vehicle gets to your service bay).