Description |
1 online resource (PDF file, 86 p.) |
Note |
"Submitted to the Graduate Faculty as partial fulfillment of the requirements for the Master of Science Degree in Electrical Engineering." |
|
"A Thesis entitled"--at head of title |
|
Title from title page of PDF document |
|
"August 2014"--title page |
|
Thesis (M.S.)--University of Toledo, 2014 |
|
Bibliography: p. 67-76 |
|
Available online via OhioLINK ETD Center |
Summary |
Global Positioning System (GPS) and Inertial Navigation System (INS) are two salient technologies delivering vehicles position, velocity, and attitude parameters for land vehicle navigation. GPS provides absolute and accurate navigation parameters over extended periods of time. However, standalone GPS perfomance deteriorates in certain scenarios such as, when a vehicle passes through urban areas or the rough forests leading to satellite signal blockages and multipath effects. Whereas, INS is a self-contained navigation technology, capable of providing navigation solution by continuously measuring linear accelerations and angular velocities in three orthogonal directions. However, depending upon INS grade, their standalone accuracy varies, due to several reasons like sensor errors, scale-factor errors, noises, and drifts. Low-cost INS consisting of MEMS sensors are being used practically due to several advantages. For instance, they are cost-effective, small in size, and light in weight. Thus, to overcome the limitations of standalone GPS and INS, and integrated INS/GPS system is required for continuous, accurate, and reliable navigation solution. In an integrated system GPS aids INS in its error modeling process thereby imporoving its long-term accuracy. On the other hand, INS bridges GPS gaps and assists in signal acquisition and reacquisition thus reducing the time and search domain required for detecting and correcting GPS cycle slips. Thus for an improved, reliable, and continuous navigation, their synergistic combination is preferred while simultaneoulsy overcoming the individual unit drawbacks. This thesis aims at developing novel statistical learning algorithms, namley Random Forest Regression, hybrid of Principal Component Regressin and Random Forest Regrssion and Quantile Retression Forests, for INS and GPS data fusion. The performance of the proposed techniques is evaluated using real field test data. The test results demonstrated the improved positioning accuracy and reduced positional drift in comparison to existing techniques during GPS outages. Through experimental demonstration, the Quantile Regression Forests has shown improved performance by providing a maximum of 87% improvement in prediction accuracy in comparison to conventional Artificial Neural Networks |
Note |
System requirements: Internet connectivity; World Wide Web browser; PDF viewer |
|
Mode of access: World Wide Web |
Subjects |
Electrical engineering
|
|
Global Positioning System
|
|
Inertial Navigation System
|
|
Quantile regression
|
|
Regression analysis
|
OCLC # |
897816429 |
|