An adaptive GLR estimator for state estimation of a maneuvering target

Abstract

This paper presents a novel adaptive generalized likelihood ratio (A-GLR) state estimator applied to rapidly maneuvering targets in pursuit-evasion scenarios. The A-GLR estimator employs a bank of adaptive models which is constructed and updated on-line. The state estimate is a probabilistic mixture of the model-matched estimates. The adaptation of the models, the model-matched estimates, and the a-posteriori probabilities of the models are calculated recursively by employing a previously developed adaptive-H 0 GLR algorithm. Numerical simulations of a maneuvering target (a ballistic missile) show that the A-GLR estimator delivers state estimates characterized by a smaller average error and a smaller covariance as compared with those obtained using the interacting multiple model (IMM) estimator.

Decision-Directed Adaptive Estimation and Guidance for an Interception Endgame

“…The adaptive-H 0 GLR detector with 70 hypotheses has moderate computational requirements, similar to those of an IMM estimator with nine models (see Ref. 28). …”