The Closed-loop Sensitivity optimization is a trajectory generation method that allows a dynamic system to best perform a task when the parameters are not perfectly known. This method minimizes the sensitivity of the state and control with regard to model uncertainties, and thus generates a more robust trajectory.
For now, this method is only developed in off-line version. The best trajectory is computed in advance for a time independent environment. The objective of this work is to extend this optimization to an online setting. In this way, the dynamic behavior could be more robust to temporal variations of the environment. The support used for this work is a Flying robot fully actuated by six tilted propellers.
References :
https://homepages.laas.fr/afranchi/robotics/?q=node/552