A filtering method of gyroscope random drift for Miniature Unmanned Helicopter

Low-cost MEMS gyroscopes used in the Miniature Unmanned Helicopter (MUH) have great random drift. In order to improve the performance of MEMS gyroscopes, a one-order AR model was established for the random drift. Then Sage-Husa adaptive Kalman filter was applied to process the random drift signal. Experiments were carried out to verify the validity of the method. Tests results demonstrate that the method based on AR model and Sage-Husa adaptive Kalman filter is convenient and effective and it significantly reduces random drift. Compared to conventional Kalman filter, Sage-Husa adaptive Kalman filter can estimate statistic characteristics of system noise and measurement noise and modify filter parameters on-time. It improves stability and adaptability and thus can give a more accurate filtering result.