Robust Model Predictive Control of Unmanned Aerial Vehicles Using Waysets

This paper introduces a new formulation of model predictive control for robust trajectory guidance of unmanned aerial vehicles. It generalizes the ubiquitous concept of waypoints to waysets, in order to provide robustness to bounded state disturbances in the presence of obstacles. Using a variable horizon formulation of model predictive control, it shows how wayset guidance combined with constraint tightening can guarantee robust recursive feasibility and finite-time completion of a control maneuver. Simulations on a point mass fixed-wing unmanned aerial vehicle model moving through a field of obstacles with wind disturbances demonstrate significant computational benefits from using waysets when compared to existing mixed-integer optimization methods that employ long prediction horizons. Using the controller's robustness to mitigate linearization error, an additional example implements the strategy on a simulated quadrotor, demonstrating how waysets can be used to control more complex nonlinear systems.