GPS/INS Integration for Autonomous Mini and Micro Aerial Vehicle Navigation Considering Time Correlated Sensor Noise

Tim Martin, Stefan Winkler, and Peter Vorsmann

Abstract:	This paper describes a method to consider the vehicle vibrations and thereby enhancing the performance of integrated navigation systems for Micro Aerial Vehicles. Core of the integrated navigation system for MAVs is a tightly-coupled GPS/INS navigation filter. A Kalman Filter relies on the correct definition of the process noise covariance matrix, which is generally defined a priori and remains fixed. The process noise mainly corresponds to the inertial sensor noise. The Kalman filter equations are based on a white process noise. But vibrations enter the inertial sensor data as nonwhite noise. Furthermore, the vibrations and, hence, the noises depend on the scenario. In the paper it is shown that the Kalman filter can be extended by an autoregressive (AR) model to consider vibration-induced noise without extending the number of system states. An adaptive technique directly provides the required stochastic properties for the AR model form inertial data. This adaptive method exposes an improved navigation solution of attitude, velocity and position, proven with experimental data.
Published in:	Proceedings of the 63rd Annual Meeting of The Institute of Navigation (2007) April 23 - 25, 2007 Royal Sonesta Hotel Cambridge, MA
Pages:	710 - 715
Cite this article:	Martin, Tim, Winkler, Stefan, Vorsmann, Peter, "GPS/INS Integration for Autonomous Mini and Micro Aerial Vehicle Navigation Considering Time Correlated Sensor Noise," Proceedings of the 63rd Annual Meeting of The Institute of Navigation (2007), Cambridge, MA, April 2007, pp. 710-715.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In