Two-stage high-order volumetric information filtering method

Abstract

The invention discloses a two-stage high-order volume information filtering method, which includes: when performing multi-sensor fusion, the volume Kalman filter is embedded into the framework of the extended information filtering algorithm to obtain the volume information filtering algorithm; the volume information filtering algorithm is based on The fifth-order Spherical-Radial volume rule is used to obtain the high-order volume information filtering algorithm; according to the characteristics of the high-order volume information filtering algorithm, the two-stage extended Kalman filter framework cannot be directly applied when it is carried out in two stages, and the dynamic matrix block method is introduced to obtain Two-stage high-order volumetric information filtering method; the present invention proposes a two-stage high-order volumetric information filtering method, the algorithm is easy to initialize, the amount of calculation is small, and the equivalent relationship between the inverse of the covariance matrix and the information matrix is ​​directly used Participating in the process of filter recursion reduces the calculation of the filter gain matrix.

Description

technical field [0001] The present invention relates to the technical field of information processing, in particular to a two-stage high-order volume information filtering method. Background technique [0002] The nonlinear filtering and estimation method based on the Bayesian framework is a research hotspot in recent years, including the extended Kalman filter (EKF) for nonlinear function approximation, the unscented Kalman filter (UKF) for Gaussian probability density approximation, and the volumetric Kalman filter. Filtering (CKF), etc.; compared with EKF, UKF and CKF do not need to calculate the Jacobian matrix and have higher estimation accuracy; CKF uses the third-order Spherical-Radial volume rule to approximate the Gaussian weighted integral, and the volume point weights used are the same and are all positive , better stability than UKF. [0003] There is following shortcoming in prior art: [0004] However, CKF also has defects, such as the estimation accuracy is ...

AI Technical Summary

Problems solved by technology

[0004] However, CKF also has defects, such as the limited estimation accuracy, the Gaussian weight integral of some simple polynomial functions cannot be accurately calculated, etc.; Filtering faces the problem of increased computational complexity, especially when the observation dimension is higher than the state dimension, which may even lead to errors in the computer's "overflow" termination during the filtering process

Images