Date of Award


Document Type


Degree Name

Master of Science in Electrical Engineering


Department of Electrical and Computer Engineering

First Advisor

John F. Raquet, PhD


Micro-Electro-Mechanical Systems (MEMS) technology holds great promise for future navigation systems because of the reduced size and cost of MEMS inertial sensors relative to conventional devices. Current MEMS devices are much less accurate than standard inertial sensors, but they can still be useful. In this thesis, data was recorded from an inexpensive MEMS inertial measurement unit and integrated with GPS measurements using a tightly-coupled Kalman filter. The overall goal of this research is to investigate the usefulness of MEMS sensors for a small, real-time, low-cost INS/GPS integration. A golf cart was used to collect dynamic data, along with a commercial INS/GPS system to provide reference data. This data was then post-processed, and the filter's performance in the position, velocity, and attitude outputs were evaluated by comparing them to the reference system. The important system features of system alignment, bias feedback, and INS resets are described, and the filter's performance is analyzed using its estimate and covariance outputs and comparing them to the true error. Filter residuals are also shown and discussed. The final results show that, with adequate processing available, the INS/GPS filter using the MEMS instruments provides good position, velocity, and attitude results over a period of up to 15 minutes, as long as the data is at least somewhat dynamic. Without vehicle motion, the vehicle yaw state tends to wander excessively, due to the bias and noise of the MEMS gyroscopes. Over a long static period, the filter's position outputs would most likely diverge and become unstable. Recommendations are made to combat this problem, among them to conduct more characterization of the MEMS sensors, and to add GPS velocity measurements as an input to the filter.

AFIT Designator


DTIC Accession Number