Autonomous Drone Landing on a Moving Naval Base using Vision-Based Robot Control

This paper proposes a novel technique of an Autonomous Drone landing on top of a moving Naval Vessel Base using Vision-Based Robot Control. To achieve an effective landing a camera is fixed under the drone which captures the images and detects the movement of the landing base. The landing base on top of the vessel is specially designed with a nested tag which is scanned by the fixed camera on the drone from a high altitude. This nested tag is a fiducial marker that is used by the imaging system of the drone as a point of reference in the Naval Vessel Base for precise landing. The controller is designed in the image coordinate frame captured by the imaging system on the drone with the help of the Jacobian Matrix. Whether the vessel remains stationary or moving, it always trembles, especially, when a vessel experiences bad weather at sea, it will be quite challenging for the drone to land on the trembling Naval Vessel Base. These unexpected movements cause the landing tag to disorient from the horizontal view of the captured frames, which makes it difficult to obtain the expected mid-point of the tag. To overcome this complication the translational and rotational motions of the drone are decoupled from the camera using the Virtual Camera Method where the mid-point of the tag is set as the feature point which helps to obtain the expected position during the horizontal plane disorientation of the tag image captured by the camera from the drone. Additionally, Kalman Filter is employed in the imaging system to reduce the observation noise so that precise estimation of the feature point can be achieved.