Auto-calibration and alignment methodology validation and characterisation for a STIM300 IMU

This thesis presents the experimental validation of an auto-calibration and auto-alignment methodology for a STIM300 IMU, using a Mitsubishi PA10 robotic arm as a controlled motion platform. The objective is to estimate sensor biases and alignment parameters in a mounted configuration, where the IMU cannot be freely repositioned after integration on its mechanical interface. The work focuses on accelerometer, inclinometer, and gyroscope biases, together with the misalignment angles between the navigation frame and the IMU frame. Particular attention is given to the coupling between yaw angle and gyroscope bias under static conditions. Since gravity-based measurements allow the estimation of pitch and roll but do not provide heading information, yaw cannot be independently determined from static data alone. For this reason, controlled rotational motion is introduced to provide additional information and separate bias-related effects from yaw-related misalignment effects. The experimental activity includes a laboratory calibration phase, used to obtain reference bias values and interface-to-IMU misalignment angles, and a real-case procedure, where the IMU is treated as already mounted and only limited static and rotational datasets are used. The results show that the proposed procedure can estimate the main calibration and alignment quantities from real experimental data. However, yaw estimation and the Y-channel gyroscope bias are highly sensitive to the accuracy of the interface-to-IMU misalignment angles alpha and beta. A sensitivity analysis confirms that small errors in these angles can introduce significant residual contributions when the robot angular velocity is removed from the gyroscope outputs. Finally, a linear same-dataset estimation of alpha and beta is investigated as a possible improvement toward a more autonomous calibration and alignment procedure.