Vision-Based Landing of Fixed-Wing Miniature Air Vehicles

This paper outlines a method for using vision-based feedback to accurately land a MAV on a visually identifiable target of approximately known location. The method presented is robust to wind, capable of handling both stationary and moving targets, and capable of cor- recting for camera misalignment, state estimation biases, and parameter estimation biases. Landing results from actual flight tests are presented which demonstrate the effectiveness of the proposed method.