Thesis (M.S.)--University of California, Davis, 2013
With the recent advancement of Micro-Electro-Mechanical Systems (MEMS) technology, inertial sensors have become easily attainable and have infiltrated a wide swath of applications. Due to the nature of the technology at its current state, the performance and reliability of these low-cost systems can leave something to be desired. As such, extensive effort has been focused on maximizing the potential of MEMS sensors in Inertial Measurement Units (IMU) through various calibration, alignment, and configuration investigations
We first design, build, and investigate two IMU sensor configurations which have been developed through previous research efforts: the twelve-accelerometer configuration which utilizes only linear acceleration readings to extract information on all six degrees of rigid body motion, and a six-accelerometer three-gyroscope configuration with dedicated angular rate sensors. To limit time-correlated drift errors in the estimated attitude of the IMUs, an Unscented Kalman Filter (UKF) is employed to fuse the data of the inertial systems with that of a tri-axial magnetometer
Next, the two magnetometer-aided systems are subjected to various dynamic tests. Inertial and magnetic data provided by the systems are then used to extract navigation estimates which are compared against actual motions captured by dual incremental encoders. Based on dynamic tracking capabilities and long-term stability, the six-accelerometer three-gyroscope IMU is found to be the superior option between the two systems developed herein. Additionally, the effects of magnetic interference, and mitigation plans for said sources of error, are outlined and provided as a reference for future efforts
