Maneuvering target state estimation based on separate modeling with high-order polynomials

Abstract

In some target tracking scenarios, the target motion is usually subject to various constraints. For a road-constrained target, the corresponding trajectory shape is independent to its dynamic characteristics due to the one-dimensional space of road. To exploit the knowledge of independence, this paper proposes a state estimation algorithm based on the separate modeling of target trajectory shape and dynamic characteristics, in which two versions based on different polynomials are considered. The idea of a sliding window is introduced, where a unique third- or second-degree polynomial with the coefficients to be estimated is used to model the target trajectory of each window. The unknown polynomial coefficients are augmented into the base state in the one-dimensional mileage coordinates and are estimated along with the base state. At every sampling period except the initial time, the proposed estimation algorithm starts the interaction stage with the previous updated base state of each filter and the modified parameter vector. The latter is determined by a least squares (LS) technique. Simulation results show that the two versions of the proposed algorithm achieve better performance than conventional estimation algorithms with coupled modeling.