|Student Name:||Daniel Marietta, Civ|
|Thesis:||Error Characterization of Vision Aided Navigation Systems|
|Location:||ENY Conference Room(Bldg 640, Rm 353)|
|Date & Time:||02/20/2013 at 1300|
|Abstract:|| Image-aided navigation (IAN) with stochastically constrained feature tracking has been a focus of research at the AFIT Advanced Navigation Technology center for use in GPS-denied environments. While previous research has shown that the IAN algorithm developed at the ANT center is effective at estimating the navigation states with a relative minimum of error, these results were based upon simulated data or small data collections and may not be indicative of real-world performance. In fact, it appears that a Monte Carlo analysis of any IAN system based upon a large-scale data collection doe snot exist in the literature. Additionally, no work performed at the ANT center with the stochastically-constrained IAN algorithm has sufficiently addressed the issues of filter consistency and divergence. As efforts are made to improve the filter performance, it is required to first have a thorough, statistically based understanding of the filter's behavior. The goal of this work is to fulfill this requirement by characterizing the errors committed by the feature tracking, EKF-based IAN system in use at the ANT center by performing a Monte Carlo analysis of a 100 run real-world data set.