Model Predictive Control for Autonomous Shipboard Landing

Autonomous landing (or perching) presents a particularly difficult control problem because landing onto a (moving) platform is a contact sensitive problem, for which safety requirements are necessarily stringent and the time available to perform the landing maneuver is limited. Furthermore,  in turbulent sea states, the deck motion is substantial and visual/sensory cues are difficult to capture and the ship-wake-helicopter interaction is difficult to model and predict. Therefore, there is a critical need for control algorithms that can guarantee performance and safety constraints, while being robust to uncertainty in input disturbances (e.g., ship wake interactions) and model parameters. This project seeks to develop a model predictive control framework for designing algorithms that guarantee fast, safe, and precise landing of helicopters onto moving ship decks.  We aim to establish feasibility of using MPC algorithm for safe and time-optimal maneuvers onto a moving platform from a computational and analytic standpoint, and then demonstrate these algorithms on a UH-60 Black Hawk RPI-GenHel-derivative model using realistic deck motion from Systematic Characterization of the Naval Environment (SCONE) database. Finally, we will quantify the effect of using lower order (linear and nonlinear) models in the MPC design on the performance of the algorithm and the trade-off of complexity against performance.

Sponsor
Office of Naval Research