Optimal Real-Time Trajectory Control of a Pitch-Hover UAV with a Two Link Manipulator

Abstract

Rotary-wing-based Unmanned Aerial Manipulators (UAM) are gaining attention for their in-place hovering capability, holonomy in motion, and safe operation due to their redundancy. Task-oriented optimal control strategies can also further improve the flight time by generating energy-optimal coordinated motions. This paper addresses the dynamic modeling, trajectory tracking and control of an UAM that consists of an under-actuated rotary-wing UAV, with a novel pitch-hover-maneuvering capability, and a 2-DOF robotic arm. A decoupled velocity-based Model Predictive Control (MPC) strategy is proposed for tracking a trajectory in the sagittal plane of the UAM while the UAV base pitch hovers in place. Conventional PD controllers were used to generate set-point velocity screws as the inputs to the MPC. A partitioned (but complete) dynamic model of the UAM was developed and used for implementing the proposed control strategy in a simulated environment. The MPC controller takes the force/torque exerted (by the arm) on the UAV base into account and augments the desired control inputs accordingly. The proposed control strategy provides the following advantages: (1) it can be implemented in real time since a linearized dynamic model of the UAV base is used, (2) it provides a generic control structure that can be applied to different classes of UAMs.