Accuracy enhancement of integrated MEMS-IMU/GPS systems for land vehicular navigation applications

This research aims at enhancing the accuracy of land vehicular navigation systems by integrating GPS and Micro-Electro-Mechanical-System (MEMS) based inertial measurement units (IMU). This comprises improving the MEMS-based inertial output signals as well as investigating the limitations of a conventional Kalman Filtering (KF) solution for MEMS-IMU/GPS integration. These limitations are due to two main reasons. The first is that a KF suppresses the effect of inertial sensor noise using GPS-derived position and velocity as updates but within a limited band of frequency. The second reason is that a KF only works well under certain predefined dynamic models and convenient input data that fit these models, which are not attainable with the utilization of MEMS-based inertial technology. Therefore, if the GPS reference solutions are lost, the accuracy of standalone MEMS-IMU navigation will drastically degrade over time.; The Wavelet Multi-Resolution Analysis (WMRA) technique is proposed in this thesis as an efficient pre-filter for MEMS-based inertial sensors outputs. Applying this pre-filtering process successfully improves the sensors' signal-to-noise ratios, removes short-term errors mixed with motion dynamics, and provides more reliable data to the KF-based MEMS-INS/GPS integration module. The results of experimental validation show the effectiveness of the proposed WMRA method in improving the accuracy of KF estimated navigation states particularly position. Moreover, the Adaptive-Neuro-Fuzzy-inference-system (ANFIS)-based algorithm is suggested and assessed to model the variations of the MEMS sensors' performance characteristics with temperature. The focus is on modeling the gyro thermal variations since it usually dominates the attainable accuracy of INS standalone navigation. Initial results show the efficiency and precision of the proposed ANFIS modeling algorithm. Finally, a new technique augmenting the powerful ANFIS predictor with the traditional KF for improving the integrated MEMS-INS/GPS system performance is presented. The proposed augmentation is utilized either to provide direct corrections to the estimated position by KF during standalone inertial navigation or to supply estimated reference position and velocity error measurements during the absence of GPS solutions, thus keeping the functionality of the KF update engine. Initial test results show the significance of the proposed ANFIS-KF augmentation in reducing position and velocity drifts during GPS outages.