Method is presented for relaxing usual assumption in sequential Bayesian or minimum variance estimation that distributions of observation errors are known; approach used is to regard distributions as normal but with unknown variances; applying Bayesian estimation theory in multistage process yields recursive equations for estimating simultaneously system state and variances; equations are like those of Kalman filter but with additional equations adjoined to produce running estimates of unknown variances; application of method to simulated trajectory estimation for interplanetary vehicle and results obtained.
Sequential estimation of observation error variances in trajectory estimation problem
AIAA J
AIAA Journal ; 5 , n 11
1967
7 pages
Article (Journal)
English
© Metadata Copyright Elsevier B. V. All rights reserved.