In state-of-the-art UAVs, the input to the control scheme - which affects the position and orientation of the UAV - is considered to be the thrust forces of the propellers. However, the actual input to the control scheme that has any effect on the system is, in fact, the angular velocity of the rotors. A direct relation is drawn between the rotor velocity and the thrust force, which is a very crude estimation that does not allow for the most efficient control of the vehicle in cases like changes in air density, malfunctions, presence of other objects in the vicinity.
This research builds up on the findings of an alumnus of the University of Twente, E.B. (Bernard) Prakken, who explored the implementation of strain gauges in a multirotor UAV. In his research, these strain gauges are used to estimate the thrust force, which in turn are fed into the control scheme to improve the stability of the system.
Essentially, this report discusses the feasibility of using the thrust forces as control inputs to observe and control the system. The goal is to excavate empirical results regarding the observability and controllability of MATLAB and Simulink simulations of the system. The system is first linearized, its observability studied, and then the propeller thrust forces were estimated using an open-loop state estimator. The results showed that the thrust forces can be estimated accurately enough to be used as control inputs to the Model Predictive Control scheme, given that the model approximations remain true.
