Design of a model predictive controller to control UAVs

In this report a continuous time model predictive controller is applied to the control of an Unmanned Aerial Vehicle (UAV). The purpose of this work, is to see if it is possible to design a good tracking controller that is capable of simulating a highly nonlinear aircraft system. For the simulations a MATLAB Simulink model of the UAV 'Ariel' is used. The nonlinear Simulink model covers all the aircraft dynamics. The model consists of ve inputs, twelve states and twelve outputs. This model is linearized at several di erent operating conditions, for use with the model predictive controller. The model predictive controller is capable of simulating in real time, making it possible to use this in on-line applications. The Ariel model is linearized to a linear time invariant state space system. The state space model is then extended. This means the outputs of the original model are included in the state vector of the extended model. With this extended state space model, it can be shown that the selected outputs can be tracked by the controller. The controller uses an algorithm which reduces the required computational power compared to classic model predictive control. The possibility to apply constraints on the controller input, the derivative of the controller input, the states and the outputs of the system, make this model predictive controller a nonlinear system. Because in real life constraints are almost always present on any dynamical system, the presented model predictive controller has a broad application scope. The simulations show that the controller performs well, after tuning of the parameters. Applying the constraints puts some limits on the possible reference trajectories, but shows that the system is very capable of handling these controls. Also, simulation of failure of one of the controls is possible, simply by setting the maximum and minimum values to zero. The UAV controller is able to use the other three controls to compensate for the lack of the fourth, keeping the aircraft at steady ight. The application of model predictive control to aircraft control is relatively new. This report shows that using a model predictive controller works well for the simulations done in this report. This method of MPC can be used not only on aircraft systems, but in general on all types of dynamic systems, with the possibility of applying constraints. For future work, a switching controller can be designed to use the di erent linearized models at di erent operating conditions. Wrapping another controller around the model predictive controller presented here, should make it possible to let the aircraft follow a set of way-points, thus not requiring a pre-de ned reference trajectory to be able to y fully autonomously.