DARPA has long had aspirations toward a guided sniper bullet, though the intriguing EXACTO project has thus far seem to focus more on a heavy .50-cal projectile to be guided smartbomb-style by a laser dot.
There is, of course, another alternative: the use of the plethora of radio signals in which we are constantly swathed. If one knows the location of every TV, radio and cellular transmitter then a multi-frequency radio could enable triangulation - Google Maps already uses nearest-base-station to get a quick fix, and BAE Systems plc proposed something along similar lines last year. DARPA started exploring that angle in 2009 but now seems to be trying to make the tech smaller and more applicable to micro-navigation.
DARPA's research rarely produces tangible results, but even its failures spend government money pushing the boundaries of what's possible. We may not have flying cars yet, but the ones driven by robots are coming along fine and DARPA drove the creation of the internet too.
GPS isn't going away: accuracy is improving and the handful of competing satellite networks being built up at the moment will improve accuracy further by aggregation, but it is current technology while DARPA's brief is to work out what comes next.
