In 2003 (the most recent year for which accurate figures are available), Wind River Systems had 16.9 percent, followed by Microsoft with 16 percent and The MathWorks with 11 percent, Gartner data showed. No other vendor had more than 5 percent market share ââ,¬" a notable contrast to the much more commoditised PC development platform. Linux is becoming increasingly visible, in part because its open code base makes it easier to modify for specific applications by excluding un-needed code.

Reflecting that depth of competition, Microsoft's Windows Mobile for Automobiles platform (a revamped version of Windows CE/Mobile) boasts of how its interface can be customised by individual developers and OEMs, a privilege it has not been willing to extend so readily on its traditional PC platforms. Similarly, Windows Mobile for Automobiles can also be compiled in device-specific formats to reduce its overall footprint, and allows exclusion of additional applications based on criteria set by the original developer.

Microsoft has invested significant sums in promoting the Windows Automotive platform, and has been rewarded with a number of deals. Many of those have been directly with car manufacturers (it boasts partnerships with BMW, Citroën, DaimlerChrysler, Fiat, Honda, Hyandai, Mercedes, Mitsubishi, Subaru, Toyota, and Volvo), but it has also worked with some third-party providers. For instance, South Korean company NexTech released an in-dash entertainment system incorporating MP3, CD, TV and radio playback, the Carman i, based on the platform in 2002.

Such third-party navigation add-ins represent one of the more visible manifestations of independent development in the in-car market. In Japan, Sony sells a navigation system known as XYZ, which creates a 3D rendering of upcoming streets ââ,¬" a more high-tech version of the voice directions offered on many existing GPS-based systems. (Whether looking at 3D street renditions rather than oncoming traffic makes for safer driving is a separate though pertinent question.)

Microsoft is hardly going to be left alone in this market space, especially as the next update of its CE-based platform, Windows Mobile for Automotive, isn't due for release until 2006 at the earliest, leaving a large window of opportunity for rivals such as IBM and Wind River. Challenges in controlling car systemsââ,¬"voice recognition is the obvious solution but demands expensive processing power ââ,¬"also mean that the battle is very far from over.

IBM itself acknowledges that in-car computing systems are generally built with vehicle safety and navigation rather than interoperability as primary design goals, but expects that to change over time. -When a system is already in place, the consumer is more likely to opt for convenience services," its Web site promoting in-car technologies notes.

So far, consumers have not been opting in great numbers. A report by Accenture examining the market for add-ons noted: -Consumer demand for telematics-related services proved less than enthusiastic."

Developing more open standards for such systems may prove to be an important step. IBM is one of several vendors backing the work of the Automotive Multimedia Interface Collaboration (AMI-C). This group's fairly ambitious, Java-like remit is to develop a set of standards -designed to give individual automakers the freedom to determine their own operating system and network drivers while at the same time enabling software developers to write applications that are capable of running in any make or model vehicle."

IBM is also predicting that much of the processing of data for in-car computing will actually take place elsewhere, utilising fast network connections to send information away for analysis. As well as cutting costs for hardware, such a system should also (at least theoretically) be easier to upgrade.

At the end of the day, in-car computing is not likely to make such visible changes to the world of driving as, say, desktop computing has to the world of office work. As Rinspeed's Frank M. Rinderknecht (the man with the scent-ejecting prototype) points out: -The driver and not the technology should be the focal point of a car". Achieving that goal remains a challenge for any developer.