The auto industry has long resisted open source, just as it has resisted many best practices in the software industry. But over the years, with more and more employees who brought a different culture to the car companies, a lot has changed very much. And so in 2018 we saw the first car to use Automotive Grade Linux (AGL) in the infotainment system: the Toyota Camry in the USA. The Mazda 3 followed in 2019. There are now AGL systems in the Toyota RAV4 and the Mazda CX-30. Subaru has been selling AGL infotainment since the 2020 Outback and Legacy models.
Now these are not cars that customers buy for their infotainment systems. In general, the customer hardly notices which operating system is behind the radio software. But from the manufacturer’s point of view, there is considerable potential in a car Linux. Perhaps a comparison with routers will help: Consumers are not interested in the fact that Linux is running on their device. But for the plaster router manufacturers, GNU / Linux was the building block that made such devices possible in their present variety in the first place.
Automotive Grade Linux sits as a consortium within the “Linux Foundation”, which companies from the automotive industry join as paying members. So far, Japanese corporations such as Toyota, Mazda, Renesas and Denso have set the tone in the AGL consortium, although it also includes Western car manufacturers, including Volkswagen, Bosch, Daimler and Ford. Many members have no specific products with AGL in the pipeline, but want them nevertheless help to shape the direction in which the project is going.
Membership costs money, but the code is available to everyone. It is available as read-only access on the AGL Git server. Interested hobbyists can, for example, get a build for the Raspberry Pi and play with it in the center console. More important for car manufacturers with large numbers of units: There are no license fees per unit sold, as is common elsewhere (e.g. at QNX). If the software engineers manage to keep the infotainment development costs low, this fact can quickly turn into huge sums of money in the volume range of Volkswagen or Toyota. Beyond the financial benefits, AGL aims to bring the less obvious benefits of open source to the automotive space:
- Independence from suppliers
- Easy adaptation through access to the source code
- Bug fixes are available to everyone.
- Much work has to be done only once across the consortium.
- Code must be built securely because it is open (no obfuscation). Or to put it the other way around: Open source code makes sloppiness shameful.
- Simplification of software audits (remember Toyota’s spaghetti code debacle)
- And finally: open source attracts good developers.
An open solution has the further advantage that it becomes the de facto standard more quickly in areas in which individual manufacturers cannot establish dominance because the niche is so small. Mercedes-Benz Vans, for example, has considered the next logical step in its “Parameterizable Special Module” (PSM), which connects the vehicle bus with attachments or superstructures: In the “adVANce” initiative, Daimler tested an AGL computer with access to the CAN bus , which evaluates additional sensors, controls loaded delivery drones, operates all IoT stuff according to freely programmable software logic. The possibilities for automation, remote control and data analysis are huge, and systems developed for this purpose could be reused in other commercial vehicles with such an AGL gateway computer.
What is (not) there
AGL is quite active at the moment: the community has released two operating system releases each year so far. That means little for the driver. Toyota does not roll out operating system updates in old Camrys. However, this shows manufacturers that development is progressing rapidly. AGL recently integrated access to Amazon’s Alexa and to Amazon’s AWS cloud offering. In addition, we are working on our own voice control interface. In addition to the center console, Linux should also run on the speedometer unit.
So that this works without real-time capabilities, it is virtualized there and strictly monitored from outside the virtual machine in which it runs. The work in this area mainly concerns the performance, so that Linux runs smoothly on the typically narrower hardware of the speedometer. However, not all software a car needs is in the repository. For example, there is a lack of navigation software that has to come from elsewhere. And there is no Play Store.
Google and Volkswagen
Despite all the advances and benefits, it is difficult to predict how important AGL will be in the automotive sector. The big competition comes from California: Google’s infotainment operating system Android Automotive OS has recently been introduced in production cars at Geely / Volvo / Polestar. Although some things still have a big question mark (why is it less capable than Android Auto with a smartphone attached?), The system shows great potential.
The app ecosystem in particular is unrivaled: Free developers can write mobility-relevant smartphone software and very easily offer UX variants for Android Auto and Android Automotive OS. The AGL consortium, on the other hand, does not offer a cross-manufacturer app shop. That is why even the very energetic, loud AGL CEO Dan Cauchy said in an interview: “There are only two solutions: AGL and Android.” The auto industry will not be able to avoid the Google question in the next few years.
Volkswagen wants to position itself as an alternative to Google with its own app ecosystem. Other manufacturers should join this according to Wolfsburg ideas. However, these ambitious plans have recently become just as silent as the plans to sell Volkswagen’s own software to other manufacturers. Instead, tidal waves of bugs in key models such as the Golf, ID.3 (test) and ID.4 keep Volkswagen’s engineers busy. The basic problem does not come from the software department, but from Herbert Diess’ misorganization. It is unclear when Volkswagen will really be able to play a part in the market for usable software. Before 2025? Unlikely. After 2025? Google is stuck in the car.
Although the software platform is called “VW.OS”, Volkswagen does not write its own operating system. Instead, they want to bring different operating systems and middleware under one roof with their own code putty. As an introduction to the operating systems to be used, these are already explicitly mentioned: Android and Linux. In projects like VW.OS, both systems will probably be found in peaceful coexistence in the future – together with other operating system animals from the automotive zoo.
If we extrapolate the current situation, AGL will probably occupy the area of low-priced infotainment systems first – the plaster router BS of the automotive sector. That may not sound very glorious, but if we let go for a moment from the German premium perspective with two SUVs per household and consider the world market: The mass market of simple basic devices in particular benefits from open source and the attitudes that go with it. And the better it works, the more interesting the entry for manufacturers of all price ranges will be.