Russell Lowe pointed me at the Digital Urban Blog, and in particular two (very different) mapping methods featured recently:

 

Google Building Maker

This is a functional hybrid of Google World, Google Sketchup, and photo-based 3D architectural reconstruction.

It’s brilliant to be able to tap into crowd-sourcing for models of city buildings. My concern is that

  1. no iterative refinement has been mentioned that allow improvements to be made to existing models (outside of just changing and resubmitting I suppose, but there should definitely be some kind of collaborative editing and version control/tracking to keep people from stepping on each others toes)
  2. Any model details beyond a certain level will not be easily reviewed. If someone does anything beyond what you can see from the map, you begin relying on honesty. (Possibly for further down the line. At the moment tools don’t seem sophisticated enough for this to be a problem)

 

MIT laser scanning robot helicopter

This is a brilliant robotic platform that seem to have some fairly meaty vision processing and navigation code running on it.

The movie below details the work led by Nick Roy at the Robust Robotics Group at CSAIL, MIT. It is extremely impressive, while a little scary at the same time, note the real-time mapping and 3D model production as the helicopter scans the environment.