Estimation of Gyroscope Biases in MEMS IMU Based on Differential AR Modeling Testing on Synthetic and Real Data

Low-cost micro electro mechanical systems (MEMSs) based inertial measurement unit (IMU) often exhibit complex systematic error components, which significantly degrades the positioning accuracy of inertial navigation system (INS). In this study, the differenced gyroscope measurements between epochs are used to eliminate the constant bias, and the second-order auto-regressive [AR(2)] model is constructed to estimate the dynamic and constant biases of gyroscope by Kalman filter (KF). The estimated constant bias is used to limit the error divergence of INS under kinematic conditions. Simulation and field tests are conducted to evaluate the effectiveness and performance of the proposed algorithm compared with the averaging method and fine alignment method. In the static test, the remaining part of the gyroscope measurement after compensating for biases is more in line with the Gaussian distribution. Allan variance analysis shows that the bias instability noise did not exist anymore. The results from the simulated static period found that the estimated constant bias by using the proposed algorithm is close to the true value. The simulated kinematic test indicates improved INS positioning accuracy compared to the average and fine alignment methods. Under various motion states with a 30 s global navigation satellite system (GNSS) interruption, improvements range from 3% to 6% and 6% to 16%, respectively. In field tests lasting 30-60 s during GNSS interruptions, accuracy improves by 3%-33% and 4%-24%, respectively.