Design Of Autopilot And Research Of Control Algorithm For Small UAV

The contents of this thesis are design of autopilot and research of control algorithm fora small UAV. Compared with manned aircrafts, unmanned aerial vehicle (UAV) is capableof automatic control, autonomous flight. In addition, UAV has many advantages such assmall volume, easy to launch and recovery, flexibility and low requirements to environment,strong survival skills, and so on. Therefore, UAV is able to perform dangerous tasks withlow cost, and it has been extensively developed and applied. Flight control system is thecore of UAV systems and responsible for automatic control and autonomous flight. In thisthesis, new autopilot emphasizes that makes it become experiment platform of fight control.Autopilot integrates dual-processor that makes it possible for implementation of parallelalgorithms.Firstly, this thesis introduces current UAV development of technology, and then itdescribes general structure of autopilot. Especially, an autopilot that takes ARM processorand DSP controller as core, has been implemented, which combines gyroscope,accelerometer, GPS receiver, pressure sensors, and electronic compass with it. We havecompleted data acquisition and fusion of sensors, data communication of groundmonitoring station, multichannel outputs of control signal and signal capture of PWM.Attitude control is a key technology for automatic flight, and attitude calculation is thekey point and difficulty in this field. This thesis presents two kinds of algorithms forattitude calculation: direction cosine matrix (DCM) and Quaternion algorithm. In order toensure the stability and reliability of computed results, the theory of complementary filtersis applied for calculation of attitudes. According to Test results, we demonstrateeffectiveness and stability of algorithms, and analyze static and dynamic performances ofalgorithms.For large amount of computation and high complexity of advanced control algorithms,an adaptive method based on sliding mode control is proposed. Compared with traditionalsliding mode control, adaptive method does not need to compute the boundary of noise anddisturbance, and just need to make sure that the boundary exists in theory. Adaptiveparameters could counteract the effects of noise and disturbance. Application of adaptivestrategy reduces computation and complexity of algorithm. Results of MATLAB simulationillustrate the correctness and validity of adaptive methods. Finally, this thesis summarizes the completed work, proposes directions of research forthe future work, and forecasts prospects for UAV and its autopilot.