A framework for visual return-to-home capability in GPS-denied environments

This paper addresses the problem of navigating an Unmanned Aerial Vehicle in an environment where GPS quality is degraded or lost. A method is described to return the aircraft from the GPS-degraded region to its launch point by means of visual navigation techniques. We record a string of overlapping images, or keyframes, on the outbound flight. On the return journey, a control law based on the homography between images is used to bring the aircraft back over each keyframe. A complementary filter is used to smooth the path defined by the homography control. An implementation and results in simulation are described and further work with a physical aircraft is proposed.