Event-Triggered Moving Horizon State Estimation for Discrete-Time Linear Systems Subject to Measurement Outliers.

State estimation is an essential issue in modern industry, which is used in a wide range of fields. Measurements corrupted by outliers have become an inevitable phenomenon leading to a degradation in the performance of estimators. To mitigate the negative impact of measurement outliers, the development of estimators with outlier suppression capabilities is crucial. In this paper, we propose a simple yet effective outlier suppression technique for discrete-time linear systems in the framework of moving horizon estimation (MHE) combined with the event-triggered mechanism. This is a novel attempt to integrate event-triggered outlier detection and correction mechanisms with the MHE approach. Specifically, we propose two event-based outlier detection methods that can effectively identify measurement outliers. Subsequently, two outlier correction techniques are designed according to the different detection schemes. Lastly, two MHE algorithms are proposed based on the corrected measurements. The proposed algorithms are applied to a target tracking simulation and compared with existing advanced outlier-robust estimators, demonstrating the effectiveness and superiority of the proposed algorithms. [ABSTRACT FROM AUTHOR]