Visual pose estimation for shipboard landing of autonomous parafoils

This paper outlines an investigation into the use of a simple, focal-plane imaging sensor for guidance of an autonomous parafoil system for approach and landing on a moving platform such as a ship underway. The perspective-projective transformation between an object in a three-dimensional world and an image on a two-dimensional plane is analyzed and then formulated using a homogeneous coordinate system. The estimation problem is addressed; specifically, the challenge of dealing with the out-of-frame condition due to parafoil oscillation as it approaches the target. A dual-mode Kalman estimation-scheme is proposed that suspends measurement when the target is out-of-frame, and incorporates a two-view measurement when the target reenters the frame.