If you look right under the google camera array, there seems to be a couple of LIDARs that could be used for 3D measurement ... Those are most likely used to produce a rough 3D model or point cloud that is coupled to the panorama. This is then used for calculating and drawing the circles and rectangles under the mouse that give a better sens of depth / 3D when navigating in the panorama.
I think that stuff is out of reach for most of us

, it is quiet advanced engineering. Though if you're curious about the next step in this direction I suggest you take a look at the current LIDAR scanners there are quiet a few solutions out there, but I was most amazed by the Faro Focus 3d lately ... they say it's capable to produce a 360x180 textured scan of 976000 points, scanning up to 120m away with millimeter accuracy, all within a few minutes ... Then you don't stitch photos to make panoramas, you stitch 3D panoramas (point clouds) to end up with 3D models.