Lattice Smooth Variable Structure Filter for Maneuvering Target Tracking with Model Uncertainty

This paper proposes a new lattice smooth variable structure filter (LSVSF) for maneuvering target tracking with model uncertainty. Under the assumption that the probability density function (PDF) is Gaussian-distributed, the nonlinear smooth variable structure filter (SVSF) framework is reconstructed by Bayesian theory. The optimal smooth boundary layer (OSBL) calculation form of the SVSF in a nonlinear framework is proposed. Then, based on lattice sampling methods with low computational complexity, the LSVSF algorithm is obtained. Finally, the LSVSF algorithm is verified on the maneuvering target tracking problem with model uncertainty by three scenarios: uniform motion (UM), coordinated turn (CT) motion and mixed motion (UM and CT). According to the simulation, the proposed LSVSF algorithm has superior tracking accuracy and robustness.