A simple computer vision based indoor positioning system for educational micro air vehicles

Computer vision is one of the main research fields for micro air vehicles. Therefore many teams around the globe are developing different camera systems which are directly mounted on a MAV or observe and track it from a fixed position. This paper presents a simple and cheap solution for a precise 40 gram onboard computer vision system based on a 6 gram Gumstix Overo Firestorm computer, a 23 gram infrared camera a custom adapter board including buck converter and connectors and three ground based high power infrared LED’s. It has great potential for educational applications because no fixed and calibrated camera installation around the flight area is necessary so it can easily be deployed and removed. The use of various filter algorithms makes it very reliable and the 752x480 pixel images are processed with 12 fps. The achieved accuracy is about 2% and the range was tested up to 10 m.