Autonomous Soft Landing of Unmanned Aerial Vehicles (UAVs) on Unmanned Surface Vessels (USVs) Using Model Predictive Control (MPC)

Collaboration between unmanned aerial vehicles (UAVs) and unmanned surface vessels (USVs) as resting stations is desirable to increase mission lengths and add safe locations for operations. However, landing a UAV on a USV presents a significant challenge due to wave-induced disturbances that continuously alter the landing conditions. Therefore, a precise estimate of the USV and advanced control strategies are required to ensure a safe landing maneuver. To efficiently track USV motion due to oscillatory wave-induced disturbances, estimation systems commonly rely on high-frequency measurements and inter-vehicle communication at high update rates. This work develops and implements a nonlinear Adaptive-Covariance Extended Kalman Filter (AEKF) with a Normalized Innovation Square Check for the proper estimation of wave-induced USV states under different sea descriptions. The AEKF uses a 6-DoF positioning measurement model that combines GNSS signals with visual detection measurements. In addition, a scheduled Model Predictive Control (MPC) algorithm is designed to enable the UAV to perform an autonomous landing. The MPC is designed and optimized for three distinct landing phases: Approach-Follow, Descend, and Landing. The performance of the MPC algorithm using the estimated USV states from the AEKF is evaluated through simulation studies of a UAV landing on a USV in various sea conditions.