Trajectory tracking and iterative learning on an unmanned aerial vehicle using parametrized model predictive control

A parametrization of state and input trajectories is used to approximate an infinite-horizon optimal control problem encountered in model predictive control. The resulting algorithm is discussed with respect to trajectory tracking, including the problem of generating feasible trajectories. In order to account for unmodeled repeatable disturbances an iterative learning scheme is applied, and as a result, the tracking performance can be improved over consecutive trials. The algorithm is applied to an unmanned aerial vehicle and shown to be computationally efficient, running onboard at a sampling rate of 100 Hz during the experiments.

I. Introduction

Model Predictive Control (MPC) is an established control strategy for addressing challenging control problems, often including input and state constraints, see for example [1] and references therein. It is based on the following procedure: at each time step, an optimal control problem is solved and the first portion of the resulting input trajectory is applied to the system. This yields an implicit feedback law providing robustness against disturbances and modeling errors.