So far, automotive software is dominated by highly proprietary, functionally limited software with virtually no software reuse. All this is accompanied by a patchwork of selective and isolated operating systems – a nightmare in terms of productivity and software economy. Yes, there are currently efforts to establish the open source operating Linux in the car, mostly in the infotainment space. But none seems to have as much momentum as Automotive Grade Linux (AGL).
Launched as a workgroup within the Linux Foundation, AGL is an initiative of mostly Japanese proponents such as automotive OEMs Mazda, Toyota and Honda as well as semiconductor manufacturer Renesas, tier one supplier Denso and consumer electronics giant Panasonic. Since its inception in 2012, a large number of technology and automotive companies have joined in; today, the members list includes Qualcomm, Pioneer, Mitsubishi Electric, Nissan and Ford as well as software vendor Wind River, supplier Continental and chipmakers NXP, nVidia, Texas Instruments and Intel – to name just a few. Notably absent: The European auto industry, including the Germans who otherwise see themselves as the avant-garde.
The AGL community is determined to bring the automotive software world up to the same level of productivity as the large rest of the IT world, said Cauchy in an interview with eeNews Europe. “The automotive industry has fallen behind the smartphone industry in terms of software,” he said. “Many customers and suppliers ask themselves why they should pay so much money for software, in the first place in the infotainment segment. There is very little software reuse, which makes all developments extremely expensive.”
Traditionally, tier one suppliers provide some kind of black box with some software inside to their customers, the auto makers. These “black boxes” remain in service until many years later the successor will hit the market. At that time, the hardware basis has changed several times, making it necessary to develop a completely new generation of software. “Our goal is providing a single, consistent platform for the entire car industry”, Cauchy said. This platform, running Linux as the operating system, would provide a basic functionality; OEMs would be able to implement their distinctive features on top.
Such a platform would create a win-win situation for all players involved, promised Cauchy. Plus, it would enable longer lifetime for old hardware.
But what about Genivi? Isn’t the Genivi consortium already readying such a vendor-independent platform, at least as far as the infotainment segment is concerned? There are several significant differences, said Cauchy who in his former career was member of the board at Genivi. “Genivi is a “bring-your-own-platform software”, Cauchy judged. “It is a specification for many platforms, but it is not one platform”, with one of the reasons for this diversity lying in the wide spectrum of hardware platforms to be served. “In contrast, AGL is one single software platform, downloadable now”. In contrast, Genivi is “not available to the world, not really open”.
Both Genivi and AGL aim at the infotainment domain. AGL however has wider goals. If it goes to the AGL community, the operating system will also run in domains like instrument cluster, ADAS, MOST (hence MOST proponent Microchip is member of the AGL community), navigation, and eventually automated driving. However, to get there, the development process will have to follow the rules of the ISO 26262 standard for functional safety. But this will be feasible. “We see no contradiction between the development processes established in the open source world and the strict rules of ISO 26262”, Cauchy said. “Actually, the principles of open source development lead to safer code. In the long term, every rogue code can and will be eliminated.”
If Automotive Grade Linux can be downloaded now, is it also production-ready? Not yet, admits Cauchy. By end of 2016, the OEMs involved will start productionizing the operating system internally.