Background:
Omnidirectional and fully actuated platforms allow for precise physical interaction with inclined walls in confined environments, thanks to their improved actuation capabilities compared to standard quadrotors. However, the unconventional placement of the propellers increases the interaction effects among them, which cannot be neglected in the control algorithms. While improved models of the thrust have been proposed in the literature, the inclusion of such models in the control algorithms has not yet been addressed.
Methodology:
This thesis addresses the application of (1) system identification methods to extract a data-based model of the aerodynamic effects between propellers in a multirotor, and (2) optimisation-based control to account for such effects in the control paradigm, thus achieving better tracking performances.
Research Questions:
How can experimental force/torque data be used to obtain an approximation of the aerodynamic interaction effects?
How to include the aerodynamic effects in a Model Predictive Control scheme for accurate physical interaction?