On modeling of random drift of MEMS gyroscope and design of Kalman filter

By virtue of lower price, smaller size and lighter weight than that of traditional rate sensors, the Micro-Electro-Mechanical System (MEMS) has been widely developed to use in low cost and miniaturized Inertia! Measurement Units (IMU). However, the MEMS gyroscopes usually have large drift, which limit its use. In this paper, a one-order autoregressive (AR) model is built for the residual signal of gyroscope and Kalman filter with colored noise is designed. In order to verify the validity of the model and the filter, static test, rate test and oscillating test are designed. The test results demonstrate that the performance of MEMS gyroscope is improved remarkably when the body does not move with great flexibility and the effectiveness will come down when the flexibility of the body increases. Analysis has been done to illustrate the reasons for the phenomena. So the technique should be used with caution when the body move with great flexibility. From a navigation point of view, the effects demonstrated seem to be meaningful and could be shown, to the best of our knowledge, for the first time.