Computer Aided Algorithms Based On Mathematics And Machine Learning For Integrated Gps And Ins Land Vehicle Navigation Systems PDF Download

An integrated navigation system consisting of INS and GPS is usually preferred due to the reduced dependency on GPS-only navigator in an area prone to poor signal reception or affected by multipath. The performance of the integrated system largely depends upon the quality of the Inertial Measurement Unit (IMU) and the integration methodology. Considering the restricted use of high grade IMU and their associated price, low-cost IMUs are becoming the preferred choice for civilian navigation purposes. MEMS based inertial sensors have made possible the development of civilian land vehicle navigation as it offers small size and low-cost. However, these low-cost inertial sensors possess high inherent sensor errors such as biases, drift, noises etc. As a result, the accuracy of the integrated system degrades rapidly in a GPS denied environment. Thus, an accurate in-lab calibration and modeling of inertial sensor errors become mandatory before being deployed. This dissertation introduces a Support Vector Regression (SVR) based IMU error modeling approach for improving the low-cost navigation system accuracy. A low-cost MEMS based IMU offered by cloud cap technology, Crista IMU is used to evaluate the SVR based error modeling approach effectiveness. Alternatively, the IMU derived navigation solution and GPS data is fused to output the more reliable navigation solution and model the errors in the inertial navigation solution simultaneously. This fusion and error modeling continues during the GPS signal availability. In the case of GPS outages, the developed error model is utilized to improve the integrated navigation system accuracy. Thus, in a continued effort to improve the standalone low-cost IMU derived navigation solution reliability during GPS outages, an intelligent technique utilizing neural networks and a hybrid of mathematics and support vector based fusion algorithms are proposed fusing INS and GPS data in an open and closed loop fashion. The performance of the proposed techniques and algorithm is evaluated using real field test data utilizing low-cost MEMS IMU, Crossbow IMU 300CC-100 and a Novatel OEM GPS receiver. The test results demonstrated the improved positioning accuracy in comparison to existing techniques and showed a substantial reduction in standalone Inertial Navigation System (INS) position error drift during GPS outages. Further, a feasibility of statistical based approaches consisting of Cubist, Random Forest and Support Vector Regression is evaluated for a low-cost INS and GPS integrated system. Through experimental demonstration, Random forest regression was found to be a suitable candidate for INS and GPS data fusion as it offers the least training time and ability to tuned the parameter automatically.