Tracking for the maneuvering target based on multiple model and moving horizon estimation

Abstract

This paper considers the tracking problem for the maneuvering target with the motion subject to some known physical constrains. For the target tracking problem, the moving horizon estimation (MHE) approach is firstly introduced, by which we can treat the physical constraints on the target motion as some useful knowledge. Under the MHE framework, we then adopt the multiple model (MM) method to describe different behaviours for the maneuvering target. To incorporate the MM method into MHE framework, we derive an estimation evolution formula and modify the weighting matrix update formula with considering the multiple model, then the physical constraint can be directly handled based on the evolved estimation. Based on the above procedures, the MM-MHE optimization and algorithm are finally presented for the tracking problem. Comparing with the adaptive Kalman filter (AKF) and the interacting multiple model (IMM) approaches, a better tracking performance can be achieved by applying our algorithm (especially for the physically constrained motion condition), which is demonstrated by a simple simulation example.