Car ecosystem is changing in front of our eyes – fr om tape recorders to pre-installed head units and now to In-Vehicle Infotainment (IVI) systems. So the most popular topic in auto software community is the future platform for ”infotainment systems” – as if somebody knows what it is. Let’s talk about expert opinions, starting with most common one – MeeGo. Notice that we consider MeeGo and GENIVI equivalent for the purposes of this discussion.
MeeGo pros are:
- Uniqueness – this is the only IVI solution available to 3rd party ISV. Sure, one can point to Centrafuse, but this is a very narrow-targeted aftermarket Windows application.
- Support – worldwide support by Intel and still-existing Nokia support; both these companies are very capable of doing fine job.
- BMW support for GENIVI – BMW’s reasons for doing so is a separate topic (which will be discussed later), but they’re obviously driving the whole alliance, including Harman and similar companies.
- Mindshare – “we do IVI on MeeGo” platform is understood (though may not be liked) by almost everybody
That’s it – now for cons:
- Shaky foundation – MeeGo is a Core OS, and collection of Linux and certain base components. For automotive industry, such bases component is IVI, which is a result of merge of Maemo and Moblin. Notice that the reason for Maemo and Moblin were sel ected not so much for being best in what they do, but rather for being available at the time when merge had to happen between Intel and Nokia development efforts. These development efforts were completely unrelated to automotive field, and the merge was done “quick and dirty” – this is the only reasonably explanation for the range of tools used (PHP, Ruby, Apache and almost any other technology you can imagine)
- No clear ownership – all automotive in-vehicle infotainment players declare that quality and reliability are their foremost goals. On the other hand, no one product designed by committee has ever achieved these goals – always there should be a single “buck stops here” player, that would allow free discussion, but would make decisions and assure follow through. Without such player, the development degenerates into chaos which, in my opinion, can be observed in GENIVI.
- Unified open competitive in-vehicle infotainment is completely against interests of Tier 1 manufacturers – if all manufacturers with all common, reliable and functional platform, while HMI will be proprietary to the vehicle manufacturer, what would their value add? Head units will become interchangeable commodity and can be purchased fr om Harman, Denso, Luxoft or fr om cheap generic Chinese manufacturers, since the functionality and reliability will be identical. Tier 1 manufacturers, who right now charge premium for domain expertise and experience, will have to compete on price – and they will lose this competition to generic shops. As a result, no major player (besides Intel, who doesn’t have critical mass and doesn’t have a major bet on the market) is in any hurry to invest development efforts into MeeGo, especially in doing boring staff, like improving quality and reliability. There’s no glory in such efforts, and there’s always a danger that another competitor would add a sexy feature (“We add navigation support to GENIVI!”), and will be the talk of the town without comparable effort.
- Community attitude towards MeeGo is also partly to blame – but it’s hard to blame people when they want to get out of MeeGo more than to contribute to it.
- Portability – MeeGo only runs on some Intel processors, and, with lim itations, on OMAP. After Nokia abandoned the N900xxx smartphones, it’s completely unclear whether anybody would add ARM support and why would they do so. As an anecdotal evidence, I’ve installed MeeGo IVI and MeeGo Notebook on five different Intel-based devices, and only IVI really run, and only on a virtual machine; Atheros division attempted to user Acer’s MeeGo notebook with similar unsuccessful results.
- Overall approach – the goal of GENIVI approach is to standardize OSS components for auto industry. Everybody understands that OSS without major sponsor can’t become reliable enough to achieve this goal – and the functionality, especially auto-specific, will be extremely lim ited. As a result, most of components are contributed not because they solve the problem at hand, but rather because of lack of critical mass to develop a proper solution. Sure, some products can be put together with duct tape and chicken wire, but not platforms, not the embedded ones. Now, let’s discuss driving forces behind this effort:
- Intel (obviously) – in head-to-head comparison with ARM, Atom doesn’t look head-and-shoulders above competitions, not at least until CarPC era arrives. Ability to sell chips into car industry as part of MeeGO IVI is a sufficient reason for the investment – and the reason NOT to invest in any other competing hardware platforms.
- BMW and other end customers – common platform for head unit software allows to radically decrease the price of Tier 1 products, and make head unit commodity, In addition, the whole ecosystem of ISV may come into existence, since suddenly a common open documented platform will be available.
- Aftermarket vendors – having common platform, their solutions will be similar and compatible to preinstalled ones, allowing for real competition. So, why then Harman and Visteon participate in this effort – the reason is likely to be two-fold: – pressure by vehicle manufacturers – their current products are falling behind and require major development investment – and they don’t have any functionally comparable solutions based on something that is not 10-year old QNX. What can be the expected result? Slow (or alternatively – pseudo-urgent) chaotic activities, including endless requirement collection, deep discussion of unimportant details and constant switching between priorities and other tasks that don’t require serious investment into engineering side.
To be continued