Dynamics and Control of a Hovering Quadrotor in Wind
Craig, William Stephen
Paley, Derek A
MetadataShow full item record
Quadrotor helicopters show great promise for a variety of missions in commercial, military, and recreational domains. Many of these missions require flight outdoors where quadrotors struggle, partially due to their high susceptibility to wind gusts. This dissertation addresses the problem of quadrotor flight in wind with (1) a physics-based analysis of the interaction between the wind and the quadrotor, (2) the addition of flow sensing onboard the quadrotor to measure external wind, and (3) both linear and nonlinear control development incorporating flow sensing and taking aerodynamic interactions into account. Using flow measurements in addition to traditional IMU sensing enables the quadrotor to react to the wind directly, rather than delaying until the wind affects the rigid-body dynamics as with IMU sensing alone. The aerodynamic response of a quadrotor to wind is modeled using blade flapping, which characterizes the tilt of the rotor plane a result of uneven lift on the blades. The model is validated by mounting a motor and propeller to a spherical pendulum and subjecting it to a wind gust. The blade-flapping model is utilized in a nonlinear geometric feedback-linearization controller that is built in a cascaded framework, first developing the inner-loop attitude controller, then the outer-loop position controller. The controller directly cancels the forces and moments resulting from aerodynamic disturbances using measurements from onboard flow probes, and also includes a variable-gain algorithm to address the inherent thrust limitations on the motors. A linear model and controller is also developed, using frequency-domain system-identification techniques to characterize the model, and handling-qualities-criteria based optimization to select gains. A linear model of the aerodynamic interactions, based on the blade-flapping work, provides flow-feedback capability similar to the nonlinear controller. Experimental testing is performed for each of the developed controllers, all of which show improvement through the use of flow feedback. Attitude is tested independently by mounting the quadrotor on a ball-joint, allowing for both gust and saturation testing. Gust rejection is also tested for both linear and nonlinear controllers in free flight, showing further benefits than considering attitude alone.