I agree the "homebuilt" market is worth a serious look, but at the same time one has to come up with something novel.
No good producing yet another "obvious product" which is an EFIS unit, which is basically an embedded "IBM PC", interfaced to an off the shelf AHRS module, a GPS and some airdata sources, with some software knocked up to generate the graphics. Pop over to the Friedrichshafen show next April and see that every man and his dog are exhibiting one of these already. It takes about 6 man-months to knock up the software, if you are clever. A lot of them are pretty naff implementations, however.
I think one could make a box like a G695, but more open (so one could load own georeferenced maps, load up Memory Map QCT maps, load up Oziexplorer maps, and PDFs of approach plates), and sell it for far less than Garmin are charging for theirs. One could sell that by mail order. The main challenge is where to get the map data from....
Best to look for a very simple standalone gadget. I can't think of one off hand but an example would be a panel mounted CO detector.
Actually there is one thing you could do: an electronic oxygen
demand regulator, like the MH O2D2 but
a) cheaper
b) fail-safe (battery flat -> max flow; the MH unit fails
shut 
)
c) externally powered with auto fallover to internal battery
d) 4 outputs
Being portable, you could sell it mail order.
Check for patents though. In the USA you can patent almost anything, regardless of obvious prior art.
10% please