Robust Moving-Horizon Estimation for Quasi-LPV Discrete-Time Systems

This paper deals with moving horizon estimation for a class of quasi-LPV systems. Under both the observability condition and the incremental Exponential Input Output-to-State Stability~(i-EIOSS) assumption, novel convergence conditions of the Moving Horizon Estimator~(MHE) are proposed. Such conditions guarantee exponential robust stability of the MHE based on a particular prediction step that is independent of the dynamics of the system. To relax the observability condition, a new prediction equation is integrated into the MHE scheme. This prediction equation exploits the dynamics of the system and contains an additional correction term depending on the estimation error and a design parameter to be determined to ensure the exponential robustness of the MHE. An application to vehicle motion estimation, using the kinematic model, is provided to show the validity and effusiveness of the proposed method, and to support the theoretical results.

This paper deals with moving horizon estimation for a class of quasi-LPV systems. Under both observability condition and incremental exponential input output-to-state stability (i-EIOSS) assumption, novel stability conditions of the moving horizon estimator (MHE) are proposed. Such conditions guarantee exponential robust stability of the MHE based on a particular prediction step that is independent of the dynamic of the system. An application to vehicle motion estimation, using the kinematic model, is provided to show the validity and effectiveness of the proposed method, and to support the theoretical results.