It is definitely possible, and in some ways almost trivial. But clearly nobody really cares enough about it.

First problem you have to tackle is that a map contains a lot of information. And the empeg display only has a certain amount of pixels to display it with. You can replace the escher image with a map image and see what the result is. From what I've tried it is either too cluttered, or such a small area that it becomes unusable for navigation.

Another problem is how to georeference the image, there are about as many ways to provide georeferencing information as there are sources for map images. And ofcourse everybody comes up with their own method, so I guess it would be no problem to invent your own. Given the georeferencing information you can do a simple matrix transformation to project the 3 dimensional gps coordinates into the 2 dimensional coordinate space of the image. Then you know exactly which pixel in the image should be in the center of the screen and copy the visible bytes over.