 All right. Let's switch gears a little bit to from kind of like a community flavor conversation to the industry flavor Joining us next is Dan Koshy, executive director of Automotive Great Linux and the general manager of Automotive at Linux Foundation. Dan joins us today to share on the state of the AGL Alliance moving forward Please welcome Dan Koshy Thank you All right, let's see if this clicker works Perfect Hi everyone, my name is Dan Koshy I'm the executive director of Automotive Great Linux as Nori just said and I'm here to share today the Really cool stuff that's going on in automotive and specifically in AGL So raise your hand. Please be honest. How many people do this? So I'm estimating at least 50% and I think some people are not being honest Okay, so why why why do we do this? It's because let's be frank the software in the vehicle has not kept up With the software in the mobile phone and the experience that consumers expect right and so It's so much easier to just put the phone there and use Google Maps or whatever And it's really sad because the car has a larger battery. It has a larger screen. It has better speakers You know, it's a great platform to do this kind of thing. And so we really need to bring the software On par with the rest of the industry such as mobile phones and why you know why and How did we get there? It's because the automotive industry traditionally has been quite fragmented You've had even within a single car company three four maybe five different operating systems anything from Proprietary RTOS that were built-in house to QNX to even for a while Microsoft Which is still out there in some cars and then also Linux came along and Linux was great But also it was Linux of you know different flavors of Linux From you know vendor a vendor B vendor C all with their own Versions of kernels and there are you know, there was no consolidation of software and this is why a GL was created It's to try to alleviate This challenge and this fragmentation and the goal of a GL has always been to build a single software platform For the whole industry, but not to be directly downloadable to the car Okay, sometimes people are mistaken. Oh, I'm gonna get involved with a GL. I'm gonna change my car I'm gonna download it to the car. That's not actually the goal of a GL the goal of a GL is to produce 70 to 80 percent of a starting point for production projects and The other 20 30 percent the car manufacturer along with their you know tier one and tier two suppliers They complete the system right they add their favorite speech recognition provider They add their favorite navigation provider. They add their favorite apps like Spotify and you know Different music apps, etc. And that's really the goal of a GL is to provide that starting point where The kernel and the device drivers for a given piece of hardware is all common to everyone the middleware for Bluetooth You know Wi-Fi Radio telephony all these things are common to everyone and then the API is for things like navigation and speech recognition, etc Are common to everyone. These are not differentiating bits They're not what people make money on in the automotive industry and so by having the starting point and providing a common interface for everyone really Hopefully will bring the car into The modern territory of software and that's really why a GL exists We've been addressing every function in the cockpit Infotainment was the first version we released back in 2016 But since then we've added instrument cluster heads up display Telematics and we are also working on functional safety and ADAS features In fact, some companies are already using a GL for ADAS which will result in a GL someday being functionally functional safety certified We have 10 manufacturers supporting the project Most of the big ones here in Japan Honda Mazda Mitsubishi motors Toyota Suzuki We have Mercedes-Benz and Volkswagen in Germany supporting the project We have SAIC in China and we have one confidential one that we're not allowed to mention and we have Hyundai in South Korea So how is a GL doing? Well, you hear a lot about Android because they have a lot of money and But we're doing quite well actually this report was generated by a company called IHS market and You can see the gray line up there is actually Android But with no Google support. It's mostly used in China where Google is not allowed to offer Google services So people they take Android they fork it and then they're on their own. They don't get updates They don't get Google Maps. They don't get Google services. So for me, it's not really like a supported OS So out of the supported OS's AGL is the market leader today. As you can see we Significantly have a bigger market share than Android And the reason Linux is going down is because in the early days before AGL existed a lot of companies started with their own Linux base and over time they've migrated to AGL because obviously AGL is automotive specific and we provide a whole stack And so those companies are moving to AGL, which is why you're seeing the downtrend And just two weeks ago yet another Publication came out which I was really happy to see SNP SNP global Said that Android will grow from 1% to 18% that's not unexpected But that AGL will grow from 14% to 23% by 2027 so we were really happy to see this And that we're we're hanging on quite nicely Versus Android, which is our friend and our competition Okay, let's switch gears and talk about software. So When we first decided to do this automotive based Distribution and the full stack We decided to name it unified code base and the reason we named it unified code base is Precisely to send a message to the industry that we're we want to try to eliminate this fragmentation and try to consolidate everyone on one platform and It's been quite a big success actually as you can see by the marketing, you know the market share data and In case you don't know This was my colleague Walt Miner's idea. We name our releases after fish And agile albacore was the first one all the way down magic Marlin but the real reason I'm showing you this chart is that we do two releases a year every year and we kind of follow a tick-tock model because Kind of like what Intel tick-tock model where the middle of the year release tends to be about bug fixing and hardening and The first release of the year tends to be feature rich and new features because we show that release at CES in Las Vegas So it tends to have more stuff to show but maybe more bugs Our latest release is prickly pike You can find all the release notes online But a few highlights is we've updated to the latest yacht to a Kirkstone We've done a lot of work on flutter virtualization and vert IO which I'm going to talk about in a few slides here We've added a kooksa valve vehicles signaling Specification support and we've done a lot of work on instrument cluster as well So let's switch gears and talk about what's coming up in the future so you may have seen a lot of announcements from Companies like Volkswagen Mercedes Toyota, et cetera that they're building their own OS and You know as as an engineer and most many of you here are engineers and technical people They're not actually building their own OS. Okay, it's it's really marketing people that I think don't know how to use the term what they're really doing is building a service delivery platform and In in almost all cases, they're using Linux and or a GL and or components thereof And so really let's not mistake that it's really a GL based or Linux based or combinations of and other industry trends is that We're getting very very powerful SOC's now multi core SOC's big cores little cores lots of cores and Unfortunately, the software is also getting way more complex more difficult to manage a GL itself is over over a hundred million lines of code and that's just Infotainment so imagine all the other functions in the vehicle, which I will talk about in a minute and then the lines are blurring between Embedded automotive type software and IT and cloud related services, which is really back to the point I was making what these car companies are trying to build They're trying to build a service delivery platforms for the consumer in the car to be able to have a much better experience and So those lines are blurring a lot of these car companies are partnering with cloud providers, etc to provide those types of services So what is a GL doing about all of this? Well, we already had a virtualization expert group that group's been active for a long time Mostly led by Panasonic which has done great work for us And we also had a container expert groups and both these expert groups They were working on very similar things, especially when it comes to this service delivery stuff and so we recently several months ago Announced that we've formed a new expert group called software defined vehicles and SDV is led by Panasonic with participation from Volkswagen AWS arm And in fact, this is an old slide. There's a lot a lot more companies participating now and The SDV expert group meets every other Tuesday, and it's very heated discussions large attendance It's by far the hottest expert group. We've ever had inside a GL So if you're in this field and you want to participate, please join us on these calls but What will SDV do what will the evolution look like for vehicle software? And this is what I'm going to try to explain because a lot of different organizations are using the term SDV But no one really knows what it means and maybe I don't either, but I'm going to try Okay, so today we have many many different processors in the car we have separate processors for instrument cluster infotainment heads-up display and then we have literally an average car has 60 to 100 ECU controllers and Literally, this is tens of millions of lines of code distributed across all these different computers It's not easy to manage. It's not easy to upgrade Hence why the automotive industry is behind the mobile phone The trend as I mentioned is very powerful multicore SUCs So what we're seeing is the consolidation of all these functions on a handful of processors. So Instrument cluster infotainment heads-up display easily can run on a single multicore processor But you need software to do it. That's what I'm going to get to ADAS cameras sensors all of that stuff that you know Driver assistance can also also be consolidated on a single processor Controllers for other things like temperature lighting etc. Can all be consolidated But the problem is that today The software is really tightly coupled with the hardware. It's an embedded device it's a single monolithic software image and If you need to make a single change for example a single security fix a safety fix something you need to recompile the whole image and Redownload it to the car and normally it's not even over the air I mean over the air upgrades are starting to happen by some manufacturers But typically requires a technician at the dealership to do it With SDV we're completely decoupling the hardware from the software. We fully support Different hypervisors some open source and also closed source proprietary ones We don't care. It's up to the industry to decide. We support vert IO for device and hardware virtualization device support for virtual devices and a complete decoupling of the hardware and software Also today as I mentioned When you want to fix a single piece or line of code you have to upgrade the entire image with SDV Our goal is that you'll be able using containers and Containerizing certain functions and search and certain applications. You'll be able to upgrade individual components and That really is a key part of the vision of delivering this software-defined vehicle today Each car let's say by a given manufacturer From different model years has different software Why because they have different hardware and because of that tight coupling Each car has a different version of the software. Yeah, there's a lot of code that's shared but it's still a different version and it's a pain in the butt to manage and With SDV the idea is to Across different model years inside like a single car manufacturer use the same software base completely doesn't matter like it's analogous to It's basically treating the car like a server is one way to put it, but it's also analogous to Apple iOS Where you know iOS runs on four or five six generations of phones It should be the same for automotive. It's not today. I know it sounds such simple thing to do but it's not today it's not that way and we need to make it that way and Things like decoupling the hardware and the software will make this possible So some of the use cases for SDV Well, the first one an obvious one is run a GL instrument cluster side-by-side with a GL infotainment second use case Let's say your car company has decided to choose Android for infotainment. No problem You can use a GL for instrument cluster side-by-side on the same processor Another use case you can run a GL instrument cluster Infotainment and then you can run all of the driver assistance a dash you can run even a separate our toss To run that stuff in a critical safety critical environment The thing is none of this is science fiction We have been demonstrating this at events for real for years at a GL It's still not deployed in vehicles because you know the automotive industry needs to catch up But you can come to our showcase and see these demos at any of our events because we've been showing this already for many years and Another use case is cloud native a GL. So people say oh, it's just everybody's jumping on to the cloud But it's actually very useful We're expanding a GL development community options by having a GL supported on AWS Graviton and This is a cloud-based arm64 Environment and again because of vert IO we're able to have the exact same software image Run in the cloud and so you can have a developer working at Starbucks on a laptop and doing you know coding debugging Testing remotely and so you don't need physical hardware anymore. So this is quite useful and we're supporting that at a GL so to summarize A GL SD fee SDV is really about modernizing in vehicle software consolidating multiple functions like instrument cluster infotainment, etc on a single SOC Simplifying the software lifecycle the deployment the management the over-the-air updates simplifying all of that making individual components upgradeable and And having a single software platform across different vehicle types and different model years So this is my attempt to explain SDV to you Okay, another very important Development at a GL so flutter is an app and UI development toolkit its open source was originally developed by Google and Toyota decided to take flutter And make an embedded automotive version of it and then contributed all their code back to a GL So we're really thankful to Toyota for doing this. We're now the home of automotive flutter and We believe strongly that flutter will be the future for and the de facto standard for UI for automotive and The reason why this is cool also is that we know that two large German car manufacturers that I can't mention are also using flutter And so when they become public it will become the de facto standard because you'll have the largest companies in the world Supporting flutter So really exciting because we're the home of that code and we just we're really pushing for it And we want we want this to happen and so recently We completely redid our AGL user interface. This is actually the actual user interface. This is our home screen I hope the colors come up, but it's really cool. You can come see it at the showcase We want to thank ICS Literally from the day we decided to start this project to today was six weeks Completely redid the UI using flutter in six weeks. And so we're really thankful to ICS for helping us design this new UI This is the HVAC system And this is the media playback, but it's better to come see it in person Okay, so I'm running out of time so quickly We're gonna be at CES where you can see more demos. This is what our booth looks like in previous years This year we're gonna have a vehicle the 2024 Lexus TX 500 will be in the booth showing AGL based system so we're really thankful to Toyota for providing this vehicle and We will be right next to Mercedes Volkswagen Hyundai in the automotive technology area So we hope that you can come and see us there Another note, we're gonna have our AGL all-member meeting winter in February here in Tokyo at the Hilton Shinjuku So please join us for that as well The CFP is now open and we'll be closing on January 21st And then one final note the AMM summer has also been decided We will be in Berlin at the Double Tree Berlin great brand new hotel July 17 to the 19th and Stay tuned for CFP information. The website is already live and you can go check it out So with that I just want to remind everyone that AGL is strongly a code first organization we've been saying this since day one and the reason for that is we're concerned that if You're a specification organization You can have multiple vendors claim to be compliant to the spec But then you end back you end up back to my first slide, which is more fragmentation because each Version of AGL would be different And so AGL is code first the starting point for AGL projects is you go to the AGL website You download the code. That's your starting point That ensures that people at least are using a very common software base. So thank you very much for your time