At the company I consult for, GEOSYS Technology Solutions some other researchers and I have developed a rapid mapping solution which can downlink images from the UAV as it flies using radio modems, and process the images in real time to produce aerial maps. Here's the result our software produces (fully automated, and nearly in realtime) for some data taken while the UAV flew over the airfield at Princeton, B.C.
The UAV we used in Princeton to fly our sensing payload was the Thunderchild, which is owned by a partner company, Universal Wing Geophysics. They've put a lot of great effort into making an easy to use, reliable platform which can carry a variety of payloads. I think it was amazing to see an un-manned plane take off and land. Here's a short video of a successful landing (sorry for the quality, my server space is very limited and I haven't learned YouTube yet).
The test in Princeton was done to validate a technology we've been developing: Geo-Intelligent Collaborative Decision Support System for Real-Time Disaster and Emergency Management (GIDE). This system combines a UAV which carries a payload package with sensors (currently GPS/IMU/digital camera) and a radio modem, with a data processing ground station which downlinks the data and provides disaster managers a view of the affected area in real time, as the UAV passes over. To test out the payload package in Princeton, I carried it in my car for a while. That required being mounted with orange flashers and driving down the center of the runway. Unfortunately, I didn't actually fly and Doc Brown was not present, nor a flux capacitor, so I also didn't time travel.
The orthoimage mosaic above shows the area near the Princeton airfield, which we flew over. The idea for disaster management is to use the great resolution of this data (we can get close to 1cm per pixel) and the rapid processing to give disaster managers a really great idea of the situation on the ground. The interface will eventually look like a live overlay layer on an existing map which has lower resolution, but more label information. Like this Google map for example: