Online Trajectory Optimization and Tracking for Nonlinear Systems - Applied to the Reaction Wheel Pendulum

Johannes Huber, Christoph Gruber and Michael Hofbaur^*

Flatness based tracking controllers are widely used in nonlinear control applications. These controllers demand smooth reference trajectories for the flat outputs and its derivatives. In most applications it is possible to design the trajectories offline, which is usually done by model based optimization. In this work, the reference for the flatness based controller is generated online, by the use of a hierarical nonlinear control structure. A high level controller, that is a Nonlinear Model Predictive Controller (NMPC), provides optimal reference trajectories to the flatness based inner control loop and keeps the system within state and control constraints. Possible structures using the two cascaded controllers are analized in simulation and a feasible solution is validated experimentally for the reaction wheel pendulum.

Mathematics Subject Classification: 93C10 93A13 93C95

Keywords: predictive control; flatness based control;

Minisymposion: Modelling and Control of Mechatronic Systems