A Robust State Estimation Method Based on Probabilistic Zonotopes Against Uncertain Inputs and Measurement Stochastic Errors

The unknown disturbances in complex environments often cause stochastic errors and uncertain inputs in measurement instruments, especially in the fields of target tracking and autonomous navigation. However, traditional Set-based state estimation and Kalman filter approaches are inadequate for modeling the noise-mixed scenario, and the robustness of the estimation outcomes is rather poor. To solve the problem, this article presents a method for target state probabilistic interval estimation, utilizing the set membership estimation approach based on probabilistic zonotopes. Initially, the proposed transformation operator effectively conducts reasonable transformation on the mixed interference of uncertain variables and stochastic errors. Subsequently, the set membership filtering is employed to predict and update the target state. Moreover, the probability assignment is proposed to convert the state estimation into probabilistic interval estimation. Finally, the robust state estimation algorithm framework based on probabilistic zonotopes is summarized. Experiments demonstrate the effectiveness and feasibility of our proposed method.