Characterization Method for Errors of Inertial Sensor Using Improved Allan Variance Algorithm

The Inertial Navigation System (INS) can provide position, velocity and attitude information with high accuracy with a short time period. In recent years, due to its small size, light weight, low consumption of electricity, and low cost, micro-electromechanical systems (MEMS), IMUs (Inertial Measurement Units) have been increasingly popular in robotics, Unmanned Aircraft Vehicles (UAVs), and Unmanned Ground Vehicles (UGVs). MEMS IMUs combine a 3-axis accelerometer, 3-axis gyroscope, 3-axis magnetometer, thermometer, and other components into a single microchip; nevertheless, the accuracy of MEMS IMU is sometimes poor, especially, the accuracy of these information decreases rapidly with time. To accurately estimate the navigation information, it is necessary to characterize the error components of the sensors. In this paper, we present a method which can simultaneously increase the computational accuracy and computational speed of Allan variance (AVAR) method with combining the variable-time averaging method with the overlapping AVAR method, in order to take advantage of the improvement methods of AVAR proposed in the literature and overcome the shortcomings. We also present an efficient noise reduction method and verified the efficiency with parameters estimated using the AVAR method. We proved the effectiveness of this algorithm using the raw data of MPU9250 in the static position. The experimental result shows that the improved AVAR approach combining variable-time method and overlapping method is better than the others.