Improvements for Vision-based Navigation of Small, Fixed-wing Unmanned Aerial Vehicles

Investigating alternative navigation approaches for use when GPS signals are unavailable is an active area of research across the globe. In this paper we focus on the navigation of small, fixed-wing unmanned aerial vehicles (UAVs) that employ vision-based approaches combined with other measurements as a replacement for GPS. We demonstrate with flight test data that vehicle attitude information, derived from cheap, MEMS-based IMUs is sufficient to improve two different types of vision processing algorithms. Secondly, we show analytically and with flight test data that range measurements to one other vehicle with global pose is sufficient to constrain the global drift of a visual inertial odometry-based navigation solution. Further, we demonstrate that such ranging information is not needed at a fast rate; that bounding can occur using data as infrequent as 0.01Hz