Using Pictures To Determine Your Location?

Microsoft has 750 researchers worldwide engaging in skunk works-type of blue sky research, just looking for good ideas. Despite accusations that the company doesn't invent things anymore, it seems like they are actually very active. The accusation should be: "Do they invent anything useful, and do the inventions make it to market?" I imagine some probably do, but I'll leave that for Redmond to argue. A recent Reuters story covered Tuesday's TechFest Fair, where MSFT researchers showed off their latest ideas. In one idea, a lost person could get a map of their location by snapping a picture of a nearby building with a camera phone, and uploading the picture to a server. The server would do a recognition match for the building, lookup the coordinates, determine their location, and deliver a local map back to the phone. MSFT has trialed the idea by uploading millions of photos of Seattle to the server. Geez, what a waste! Somebody tell these guys that 99% of US phones are shipping with GPS today! A complicated system using photos is redundant, expensive, error-prone, and slower than the existing A-GPS.

Ironically, just before I wrote this post, my wife asked me, "Is there a way I can go into my contacts in Outlook, and click on a "show map" button for one of my contacts?" The answer is actually, "Yes" but it's buried so far in sub-menus that most people don't know about it. (click "Read More" to learn how.) It's frustrating that Microsoft sometimes makes the costly and redundant easy, while making the intuitive and useful difficult.
So you wanna look up maps from your Outlook Contacts list? Select the contact, open it (as if to edit), click on the Actions menu, and choose Display Map. That will take you to MSFT Live Search, and an AJAX-based map with some useful tools. You would think Microsoft would make it easier, since it can drive more traffic to their web properties.

BTW, I know the science of what the researchers are doing is interesting, and the "building recognition" may be useful in other ways. But the researchers should have some responsibility to relate their ideas to actual real world uses, not just technology for technology's sake. If I were funding that research, I'd want each labcoat to give me a top 3 list of the ways his idea could be realistically deployed. You wouldn't have to be right in 5 years, but it would have to make sense today. Here's a couple:

1) Public safety could match buildings to addresses, so firemen would know what the building their rushing to looks like before they get there, because it's displayed in a LCD screen in the fire truck. (the LCD could display all known data about the building). MSFT matches the address to the images, and provides the images for display.

2) For driving directions on MSN.com, they could provide a photo of the destination address, so that you can recognize it more safely than checking for small address numbers.

3) For photo organizing sites like shutterbug, etc, the MSFT service could match your photos to the locations they were taken, and automatically tag them with meta information.

OK, that's three, and I've been thinking about it for 2 minutes. I may be smart, but the MSFT labcoats are surely smarter. So get some realistic ideas, already!