A Thorough Convergence Analysis of Unscented Kalman Filter
组织者
演讲者
孙泽钜
时间
2023年06月12日 14:30 至 15:00
地点
数学系理科楼A-203
摘要
In this talk, we will give a thorough convergence analysis of unscented Kalman filter (UKF) for continuous nonlinear filtering systems. The convergence analysis is based on a comparison between the evolution of UKF and the Kushner-Stratonovich equation. We will first restrict ourselves for systems with bounded coefficients and derive some sufficient conditions that guarantee the convergence of UKF. Results illustrates that the skewness of the conditional probability distribution may be a key factor to the efficiency of Kalman-type algorithms for nonlinear filters.