Research On Unmanned Helicopter Control Technology And Application

Nowadays, the small unmanned helicopter has been studied as a platform for flight control andvisual servo control by many institutions. Base on the existing results, the small unmanned helicoptersystem overall design, helicopter modeling, visual localization system design, robust controller designwere investigated in this thesis. The main content was summarized as follows:First of all, the unmanned helicopter overall system, which include flight control system, visuallocalization system and ground station system, was presented in the thesis. The hardware devices andtheir performance were listed. The description of system states, the emergency response of the systemfailure and the communication protocols between the helicopter and the ground station were studied.Secondly, with analysis of the rotor dynamics, aerodynamic forces and moments, the nonlinearmotion model was established for the small scale helicopter. Then, based on the characteristic ofhovering and low translational speed flight dynamic, the nonlinear helicopter dynamic model waslinearized as a four inputs four outputs state-space model with unknown parameters. The unknownparameters of the state-space model were identified by using black box method. To simplify theproblem, the model was separated in four single input single output channels and each channel wasidentified respectively. And, by integrating the four channels and identifying the couple effectbetween each channel, the four inputs four outputs model was obtained.Following, a visual localization system was proposed due to the shortage of GPS in indoorlocalization. To improve the accuracy, the camera was calibrated. The affine transform between twoadjacent frames was obtained by using2D feature extraction and descriptor matching. The3Dlocalization was achieved by fusing the information of the sensors and affine transform with KalmanFilter.Finally, the uncertain model was obtained according to multiple groups of sensors data, andfrequency band wide of the system and the noise were obtained by Fourier transform of the sensordata. On this basis, a robust controller was designed by using μ method to eliminate the noiseinfluence and the control errors. The simulation results in time domain and the analyses in frequencydomain showed that the close-loop system was stable and the control requirements were achieved.