Advanced control for miniature helicopters: modelling, design and flight test

Unmanned aerial vehicles (UAV) have been receiving unprecedented development during the past two decades. Among different types of UAVs, unmanned helicopters exhibit promising features gained from vertical-takeoff-and-landing, which make them as a versatile platform for both military and civil applications. The work reported in this thesis aims to apply advanced control techniques, in particular model predictive control (MPC), to an autonomous helicopter in order to enhance its performance and capability. First, a rapid prototyping testbed is developed to enable indoor flight testing for miniature helicopters. This testbed is able to simultaneously observe the flight state, carry out complicated algorithms and realtime control of helicopters all in a Matlab/Simulink environment, which provides a streamline process from algorithm development, simulation to flight tests. Next, the modelling and system identification for small-scale helicopters are studied. A parametric model is developed and the unknown parameters are estimated through the designed identification process. After a mathematical model of the selected helicopter is available, three MPC based control algorithms are developed focusing on different aspects in the operation of autonomous helicopters. The first algorithm is a nonlinear MPC framework. A piecewise constant scheme is used in the MPC formulation to reduce the intensive computation load. A two-level framework is suggested where the nonlinear MPC is combined with a low-level linear controller to allow its application on the systems with fast dynamics. The second algorithm solves the local path planning and the successive tracking control by using nonlinear and linear MPC, respectively. The kinematics and obstacle information are incorporated in the path planning, and the linear dynamics are used to design a flight controller. A guidance compensator dynamically links the path planner and flight controller. The third algorithm focuses on the further reduction of computational load in a MPC scheme and the trajectory tracking control in the presence of uncertainties and disturbances. An explicit nonlinear MPC is developed for helicopters to avoid online optimisation, which is then integrated with a nonlinear disturbance observer to significantly improve its robustness and disturbance attenuation. All these algorithms have been verified by flight tests for autonomous helicopters in the dedicated rapid prototyping testbed developed in this thesis