Uav Automatic Landing On A Ship-Deck, Multivariate Sensor Fusion For Robust State Estimation

Abstract

In response to the growing demand for autonomy and reliability of Uncrewed Aerial Vehicles (UAV), this research presents a navigation solution for the automatic landing of an uncrewed rotorcraft on a ship-deck based on computer vision enhanced through sensor fusion with data from the attitude heading reference system (AHRS) and radio beacons. Conducted by a team of Leonardo Helicopter Division, this study addresses the case of a GPS or data-link loss and includes the creation of a relative pose estimation algorithm based on the detection of Fiducial Markers, complemented by the formulation of an Unscented Kalman Filter (UKF) and sensor modeling based on test data. Results are obtained in a simulation environment relying on ROS 2 (Robot Operative System 2) for inter-process communication, Ansys AVxcelerate for optical sensor and environment modeling, and a proprietary software responsible for guidance, control logics, and dynamics simulation. Research findings showcase promising outcomes and open avenues for future enhancements, including real-world testing.