A Study on Positioning for Industrial Tools using Inertial Sensors

Abstract: This thesis investigates the currently achievable estimates of position and orientation using inertial measurement units(IMU) intended for use in indoor positioning of tightening tools. The study first introduces the reader to the problems inherent in estimating position and orientation using inertial measurement units. The reader is then introduced to the state of the art methods in estimating position and orientation and these methods are then implemented and studied. To evaluate the effectiveness of these methods, first an IMU is modeled in Simulink and the model is then verified in the time and frequency domain. Further the model is simulated in static and dynamic conditions to simulate movements in an indoor environment. The data from these simulations is then input to 3 state of the art sensor fusion algorithms to estimate orientation. To investigate the state of the art in estimating position, a Zero Velocity Update(ZUPT) aided Kalman filter is implemented in Matlab. Real world datasets from a commercial and navigation grade IMU is collected for two trajectories which emulate a tightening operation. The position accuracy is then estimated for the entire dataset and statistics are calculated for the mean accuracy and the results are presented. The results are a significant improvement on the state of the art and indicate that an accuracy of 1 degree in Roll and Pitch and a positioning accuracy of 1m/min is achievable for this application using off the shelf commercial grade inertial sensors.