Improved MAV Attitude Estimation through Coupled Acceleration Estimation

Bryce B. Ready and Clark N. Taylor

Abstract:	As Unmanned Mini Air Vehicles (MAVs) become increasingly popular, with an ever-widening circle of commercial and military applications, MAV navigation technologies are becoming an increasingly important area of research focus. One of the critical challenges in this field is the ability to estimate MAV attitude in a robust and accurate manner. Accurate MAV attitude is critical for an ever-growing list of MAV applications, such as geo-registered video mosaicing and various target localization methods. On MAV platforms, accelerometers have long been used to provide an estimate of the direction of gravity, which in turn allows estimation of aircraft pitch and roll. On an MAV that is not accelerating, this estimate will be unbiased. However, any acceleration such as increased throttle or centripetal acceleration in a turn will be added to the accelerometer measurements, and will thus bias the estimate of pitch and roll. Several exiting methods estimate acceleration to remove this bias, using GPS measurements or airspeed measurements and various assumptions about aircraft dynamics in order to do so. We introduce a novel attitude estimator that incorporates position, velocity, acceleration, and attitude into a single Unscented Kalman Filter state. This approach takes full advantage of the cross correlations induced between these quantities that are induced by the accelerometer measurements. It is thus able to better use information available in gyro, accelerometer, and pressure measurements to correct attitude because of this tighter coupling. The structure of the Kalman Filter allows information from previous measurements (including GPS, airspeed, and accelerometer measurements) to improve pose estimate quality, even if individual prior measurements had unknown bias. If an estimate of MAV air velocity is available, this approach allows us to estimate pose without making any assumptions about aircraft sideslip and angle of attack. In the more common case that only the magnitude of air velocity is directly measurable by pressure sensors, we can still incorporate a-priori information about the direction of aircraft velocity and obtain good pose estimates. We present simulation results showing a significant improvement in attitude estimation quality as compared with current state-of-the-art methods.
Published in:	Proceedings of the 2009 International Technical Meeting of The Institute of Navigation January 26 - 28, 2009 Disney's Paradise Pier Hotel Anaheim, CA
Pages:	1054 - 1061
Cite this article:	Ready, Bryce B., Taylor, Clark N., "Improved MAV Attitude Estimation through Coupled Acceleration Estimation," Proceedings of the 2009 International Technical Meeting of The Institute of Navigation, Anaheim, CA, January 2009, pp. 1054-1061.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In