Quadrotor Controller Design Based On Model Reference Sliding Mode Control

Quadrotor is a type of unmanned aerial vehicles featured with VTOL (Vertical Take-Off and Landing), fixed-point hovering capability, high maneuverability, low cost and excellent performance. Due to the advantages, it has good practical applicability and great prospects in terms of military reconnaissance, urban transport, disaster relief and other aspects, and it has unique research value and engineering significance. Moreover, quadrotor is nonlinear, strong coupling and sensitive to disturbance, which brings difficulty to its mathematical modeling, flight attitude information acquisition and design of controllers. On the basis of previous work, a research on the key problems of the quadrotor control system has been taken, which includes the establishment of experimental platform and nonlinear modeling, the study of data fusion algorithm for attitude estimation and the design of flight control method.Firstly, by taking the demands of machinery and control as criterions, the design of airframe, airborne hardware system and software system are conducted. The airframe design is analyzed from material and structure. Moreover, the hardware system is divided into different modules including microprocessors, sensors and actuators. In addition, the software system is introduced from the task assignments and software process of two microprocessors and the data acquisition and processing of sensors.Secondly, the principles of quadrotor flight are elaborated on the basis of the definition of flight coordinate systems and attitude. According to the analysis of the structural features of the airframe, rotor dynamics and Newton-Euler dynamic equations, the six-DOF (Degree of Freedom) nonlinear model of quadrotor is established.Thirdly, multi-sensor data fusion algorithm based on quaternion and Extended Kalman Filter is designed for attitude estimation. After analyzing the characteristics of multi-sensor data, quadrotor attitude is obtained by the integration of the gyroscope, accelerometer and magnetometer data. Simulation results demonstrate the effectiveness of the proposed data fusion algorithm.Finally, the decoupled linear model is obtained through linearization and parameter identification of the six-DOF nonlinear model with reasonable simplifications. Attitude controllers and altitude controllers, based on model reference sliding mode control, are designed for the above linear model. Simulation results corroborate the performance of the proposed controllers.