Simulated Autonomous Landing of a Rotorcraft Unmanned Aerial Vehicle in a Non-cooperative Environment

This paper describes a multi-stage process that should allow a Rotorcraft Unmanned Aerial Vehicle (RUAV) to land autonomously at a non-cooperative site using vision-based navigation in the absence of a conventional navigation sensor, such as the Global Positioning System. A non-cooperative site has no prearranged navigational aids that would help guide the RUAV to a predetermined safe landing site. Machine vision, using stereoscopic imagery, is used to first locate a safe landing area in an obstacle field. A single camera is then used to track features on the ground to guide the vehicle to the safe landing area. Descriptions of the algorithms, along with verification results obtained using full-mission simulation, are presented. The outcome of the research shows that the algorithms perform adequately in simulation, and currently developed portions of the hardware integration perform equally well.