Continuous time, adaptive, ML based Kalman filtering is considered. First, the geometric structure of the MLE limit parameter sets is characterized. Then, based on the geometric study, a recursive adaptive filtering scheme is derived to obtain asymptotically optimal (minimum variance) state estimates.

D. Levanony, A. Baghi (Dept. of Applied Mathematics, University of Twente, The Netherlands)

Mon Nov 13 10:43:02 EST 1995