Maneuvering target tracking with unknown acceleration using retrospective-cost-based adaptive input and state estimation

Abstract

In this paper, we apply retrospective-cost-based adaptive input and state estimation (RCAISE) to maneuvering target tracking. Conventional methods assume that the maneuvering process is a random process. In contrast, RCAISE uses an adaptive input estimator to estimate the unknown maneuvering acceleration. This estimator optimizes the retrospective performance to drive the estimated maneuvering acceleration to approximate the actual maneuvering acceleration. Using the maneuvering target tracking model, a state estimator is constructed to provide the optimal estimate of the state. Numerical simulations illustrate the effectiveness and feasibility of RCAISE with comparison to the conventional methods.