DIVIDED DIFFERENCE FILTER-BASED DATA FUSION ALGORITHMS FOR ATTITUDE ESTIMATION

This paper presents estimation of rigid body orientation using the newly developed divided difference filters (DDF). These estimators utilize considerable advantages of using polynomial approximation obtained with an interpolation formula instead of Taylor series approximation used in extended Kalman filter (EKF) and its higher order relatives. This accommodates easy implementation of the filters, and it enables state estimation even when there are singular points in which the derivatives are undefined. Attitude estimation systems often use two or more different sensors to obtain reliable data. Regarding this fact the testbed contained rate gyros, gravimetricinclinometers and magnetic compass. The suggested approach is that data from these sensors are fused in order to achieve the excellent dynamic response of an inertial orientation estimation system without drift, and without the acceleration sensitivity of inclinometers. Also, because of using gyro modeling instead of dynamic modeling theproposed implementation is independent of the structure of the platform and can easily be transferred to different platforms, which carry an equivalent set of sensors. Extensive testing of the filter with actual sensor data proved it to be satisfactory.

Keywords:

Attitude Estimation , Divided Difference Filter (DDF) , Extended Kalman Filter (EKF) , Data Fusion , Strapdown Inertial Navigation System (INS