Model-Scale Evaluation of Autonomous Ship Landing Guidance and Control Modes

This paper presents the results of an extensive model-scale experimental evaluation of autonomous ship landing guidance and control modes, with flight tests performed in the Maneuvering and Seakeeping (MASK) Basin at the U.S. Naval Surface Warfare Center Carderock Division. The experiments were performed using a commodities-based multirotor unmanned aerial vehicle (UAV) operating from a 20-ft-longmodel scale ship subject to scaled wave conditions. During testing, two separate guidance algorithms were evaluated: a quadratic programming (QP) based landing algorithm that plans the trajectory to a forecasted deck state and a simpler "baseline"method that tracks deck motions while closing the distance between the aircraft and deck at a constant rate. Both algorithms commanded a Froude-scaled explicit model following control law, and the control law parameters were modified to progressively degrade aircraft tracking bandwidths. The results showed that the predictive capabilities of the QP algorithm allowed more direct landing paths to be planned when compared to the baseline algorithm and also allowed the QP algorithm to land with lower tracking bandwidths. But while the QP algorithm performed well in the majority of cases, there were several landings where a combination of poor deck state predictions and how the QP algorithm utilizes predictions to choose a land time resulted in significant terminal velocity and attitude errors. The baseline guidance algorithm, on the other hand, proved to be both simple and reliable when the UAV was in high bandwidth configurations but required a high reference tracking bandwidth.