Data fusion for accurate UAV navigation

Abstract

The work in this thesis is focused on selecting, implementing and evaluating a number of data fusion techniques which are suitable for UAV navigation. The work presents a selection of most commonly used sensors for navigation purposes, together with a number of data fusion techniques that are based on the Kalman Filter. The use of data fusion techniques allows the Kalman Filter to utilize information from multiple sensors to improve the filter output accuracy. A test bench involving a turntable was designed and constructed, so that the data fusion techniques could be tested in a controlled and repeatable environment. This allowed a number of sensors to be placed strategically on the moving turntable, thereby emulating a UAV turning at different angular velocities. Using the test bench, a number of experiments were conducted, where the turntable was rotated at different constant angular velocities, and the outputs of the fusion methods were analysed and compared. Finally, an experiment was conducted where the turntable was given a specific angular velocity profile, to see how each method responds to different changes in angular velocity. The results obtained indicate the capability extent of such data fusion techniques in reducing the uncertainty of typical state estimates required for navigation purposes.