Aeromodel Telemetry through Image Recognition

This is a quite old project, but back in 2009 it was cool, so in 2010 I published a small poster in our local student symposium SIITME

So, my good friend John, a crazy aeromodel builder and FPV pilot wanted a fancy feature on his aeromodel: to be able to see it position in Google Earth without having to buy a telemetry system. The guy already had FPV with a OSD from “Intelligent flight”.

Fig 1: PAL image with OSD

An low cost USB frame grabber was used to acquire the pal image in LabVIEW. We relied on LabVIEW IMAQ libraries for extracting information from the image. Back then in 2009 OpenCV was kind of new, so the program was development in Labview, since it’s fast.

Fig 2: Image acquired in LabVIEW and displayed as intensity Chart

From the acquired image we new the position of GPS coordinates so we had to recognize the numbers from the image and to recreate a new NMEA string as if the GPS is attached to the pilots computer but with data from the aeromodel. Especially annoying was to not mix 8 with 0 in conditions of high image noise but after properly training the algorithm the situation improved.

Fig 3 Reconstructed NMEA string

The Reconstructed NMEA string was supplied through a virtual serial port in a third party program: Earth Bridge and thus we got the position of the aeromodel in Google earth

Fig 4: Earth Bridge

Fig 5: Position reported in Google Earth

Design a site like this with WordPress.com
Get started