In-situ Gyroscope Calibration Based on Accelerometer Data

The paper presents a novel calibration method for gyroscopes and accelerometers. Contrary to existing methods the proposed one does not require a rotating table or other special equipment. To perform the calibration a user needs to make a series of sequential rotations of inertial measurement unit (IMU) separated by standstills. To find the sensor errors the cost function is defined in terms of orientation differences between accelerometer and gyroscope reported orientations. Then this function is minimized with respect to calibration parameters, that include scale factors, axis non-orthogonalities, biases and misalignment between gyroscope and accelerometer triads. The proposed method has been verified through Monte-Carlo simulations using synthesized IMU data. Besides the method was tested on real data from MPU-9250 sensors. In both cases, the method was proved to properly find calibration parameters. The simulations revealed that the differences between true and estimated sensor error parameters were less than 0.1% of their true value. The experiments using real and simulated data showed the significant elimination of orientation error after calibration. Moreover, the contribution of gyroscope scale and non-orthogonality errors to the total orientation error was estimated. The method implementation in Python together with the inertial data simulator and real sensor data are provided publicly**Reproducible research: all files and software for data processing used in experiments and simulations are available under an open-source license at https://github.com/mikoff/imu-calib..