A constrained robust switching MPC structure for tilt-rotor UAV trajectory tracking problem

Abstract

In tilt-rotor UAVs, both the fuselage and tilting rotors contribute to the vehicle's rotational motion. Consequently, the system's dynamics rise to a highly-nonlinear system, making it challenging to find feasible and desired control solutions. The common control practices devise a logic-based controller to switch between different flight modes or map the control inputs to the conventional helicopter-type control inputs. However, they fail to provide energy-efficient fast trajectory tracking, especially in the presence of external disturbances. This paper proposes a general-model dynamic formulation and a two-layered constrained Model Predictive Control (MPC) strategy to tackle the trajectory tracking problem for tilt-rotor UAVs. After splitting the vehicle's dynamics into translational and rotational parts, a constrained Linear MPC (LMPC) is designed for the translational dynamic to follow a reference trajectory. We formulate the LMPC as a Quadratically-Constrained Quadratic Problem (QCQP) that leads to a feasible set-point solution for the rotational control layer without violating the physical constraints. Also, an optimizer is designed to generate a thrust vector, which leverages the vehicle's