Distributionally Robust State Estimation for Highly Maneuvering Target Tracking With Model Uncertainty and Impulsive Measurement Outliers

Due to the high maneuverability of the non-cooperative target and the complexity of the confrontation environment, the uncertainty of the tracking model and impulsive outliers in measurements degrade tracking accuracy and may even lead to complete loss of tracking. Existing research can hardly address these challenges. To ensure accurate tracking in the presence of uncertainty and impulsive measurement outliers (IMOs), we propose a distributionally robust state estimation (DRSE) method based on moment-based ambiguity sets. First, virtual maneuvering noise and a first-order Markov process are utilized to describe the maneuvering acceleration, while a set of independent and identically distributed random variables is used to characterize the interval length of IMO, thus constructing the process and measurement models, respectively. Then, the uncertainty of model is represented by moment-based ambiguity sets, and the state is estimated under the worst case conditional prior distribution. Furthermore, we employ an adaptive saturation mechanism to mitigate the impact of IMO, thereby ensuring robust-bounded-error state estimation in the presence of outliers. Finally, a glide trajectory of a typical hypersonic vehicle is established in this study. The numerical experiment results demonstrate the algorithm's effective handling of tracking model uncertainty and IMO.