Trajectory Tracking Control Based On Iterative Learning For Unmanned Aerial Vehicle

The quadrotor originates in 1907, in the early time all aspects of technologies are not mature enough to have a relatively quick development. In recent years, with the progress of aerial robot related technologies, as well as its own unique advantages, four-rotor helicopter has a great value of application and a lot of broad prospects for development.However, since the quadrotor is a nonlinear and strong coupling system, these properties make the design of its flight control system to be more complex. Because the development of quadrotor gradually tend to be miniaturization, so the inertial sensors which to be used by control system are lower power consumption of MEMS sensors. Due to the reason of the manufacturing process, there are some errors existing in these sensors, so in addition to the need for design a good control algorithm, the analyses and calibration of sensor and multi-sensor data fusion is particularly important. These two aspects of the key issues are researched.In this dissertation, the four-rotor helicopters of flight principle as well as some basic knowledge which is used in the follow-up work are introduced firstly,The four rotor dynamics model is established accordance with the relevant knowledge. The repetitive disturbances and unmolded errors of the system are estimated with iterative learning control algorithm. The algorithm is used based on dynamical model of four rotor helicopter in combination with the optimal estimation characteristics of Kalman filtering. Optimal iterative learning algorithm is selected based on function within discrete quadratic equations. The algorithm compensates the estimated model error and is assessed with Matlab. The results of various trajectory tracking will be showed after simulation, it demonstrates effectiveness of the iterative learning control algorithm.In this dissertation, hardware experimental platform of the four-rotor flight control system is established and sensors in the platform are calibrated and filtered. Then a large number of experimental has been completed to verify the overall filtering performance. In addition, chi-square test methods is used to deal with fault detection and isolation, the data results showed that the effectiveness of the method.