Gain scheduled robust control of unmanned aerial vehicles

In this thesis, gain scheduled robust control of unmanned aerial vehicles (UAV) is examined. Mathematical H_∞ optimization problem is solved using Linear Matrix Inequalties (LMIs) to synthesize sub-optimal controllers. Mixed sensitivity H_∞ optimization and H_∞ loop shaping methods are used in the controller design. First, a nonlinear helicopter model is built, trimmed and linearized from which an approximate affine-parameter-dependent model is constructed. Then, LMI based mixed sensivity H_∞ controllers are designed in order to achieve stabilization and reference tracking for a small unmanned helicopter at various flight conditions. Local H_∞ controllers are designed at trim conditions based on local linear models. The pointwise controllers achieve local stability and performance, but fail in stabilization and tracking over the full envelope. A scheduling controller is built by blending the local controller outputs. In addition, local H_∞ controllers are designed with common Lyapunov function. This allows controller scheduling between the design points with guaranteed stability and performance across the design envelope. Moreover, a parameter-dependent controller is synthesized to stabilize the affine-parameter-dependent helicopter model attaining stabilization and reference tracking for the family of linear models. All methods except for the local controller approach yield in satisfactory performance over the full flight envelope. Afterwards, H_∞ loop-shaping controller design is studied. Synthesized controler are scheduled parametrically with guaranteed robust stability over multiple operating points. Synthesis LMIs for state feedback and dynamic output feedback are derived in the parametric H_∞ loop-shaping framework. The results are extended to parameter-dependent plants to build a parameter-dependent controller utilizing a common Lyapunov function. The developed theory is applied to a small helicopter model, for which the operating region is covered by a family of linear models at a grid of operating points. It is shown through linear and nonlinear simulations that a desired loop shape is attained by the parameter dependent controller. Satisfactory tracking is achieved and stability is retained, even under mass and inertia variations. Moreover, to show the generality of the proposed technique, an autopilot is designed for an UAV where one of the lateral control surfaces, i.e. the aileron, becomes jammed and unusable. The autopilot handles the automatic recovery, autonomous guidance and landing of the disabled UAV. An accurate nonlinear aircraft model is used to build flight control laws for the UAV using loop-shaping theory to decouple longitudinal and lateral channels. The designed autopilot is tested on an example distress scenario involving aileron servoactuator jam. It is confirmed through hardware-in-the-loop (HIL) simulations that the autopilot design is capable of resuming safe flight and autonomous navigation under the fault scenario and is able to safely land the UAV to a target runway. Keywords: H_∞ Optimization, Mixed sensitivity, Loop shaping, Robust control, Paramater dependent control, Modelling, Simulation.