Improved Performance of Recursive Tracking Filters Using Batch Initialization and Process Noise Adaptation
Michael E. Hough · Journal of Guidance Control and Dynamics · 1999
Performance of nonlinear recursive tracking filters may be improved using batch initialization and process noise adaptation. These techniques are applied to the problem of radar tracking of a non- maneuvering target vehicle in exoatmospheric flight. Although this problem has received considerable attention in the estimation literature, the approach discussed in this article is new in two respects. The novel features include the method of covariance initialization of the recursive filter and an adaptive model for a process-noise matrix. A batch initialization algorithm generates a covariance matrix with non-zero correlations of position and velocity errors, and this feature accelerates the convergence of errors in the estimates of velocity. The process-noise matrix is based on the gravity-gradient effect, and the tuning method is adaptive because it uses the most recent state estimate and covariance matrix. This feature compensates for nonlinearities in the filter dynamics model, and it improves the agreement between filter covariance and actual errors in the estimates.