In the past week or so, there have been a number of posts and articles in the geo-blogosphere and "tweets" in the geo-twittersphere about Augmented Reality. These have been about a new set of mobile consumer applications for devices such as the iPhone.

A note of caution against all this hype, in sounded in the article "Extra layer of reality gets off to a shaky start" in the current edition of New Scientist (19th Sept 2009, pp.20-21). Problems listed include the lack of sufficiently accurate GPS devices. New Scientist quote an error of up to 70m with an iPhone in Downtown San Francisco, and the compass having typical magnetic-compass problems (ie. 180 degree deviations near metal sculptures).  The size of the GPS device is a problem - I have seen reports that the latest generation of GPS devices that are the size of a thumb drive being 'good' in ideal conditions but satellite lock and distance errors are notably inferior in poor environments such as urban canyons. Even the Trimble Field Computer we use for the Costa Rica field work can only manage 6m accuracy without differential correction, and this is a professional device with a special built-in antenna and is larger than an 80s-era cell phone!

The article notes that miniature GPS devices will probably improve, but will never be good enough. To achieve the sub-metre accuracy that many augmented reality applications require, will require image recognition. Such a device would use GPS and the compass as a rough estimate of position, and then the device could use image analysis to try and match the view with a digital 3d model. Occipital (Boulder,CO) are quoted as saying they have software which can do exactly this, and give locations accurate to within 1m. 

The article ends by voicing the fear that consumers will try these initial augmented reality applications and find they do not live up to the hyped expectations, decide the technology is a dud, and stay away. Parallels are drawn with the hyping of virtual reality in the early 1990s.