ABSTRACT
Information fusion filter theory is an important branch of multisensory information fusion, is mainly focused on the study of multisensory information fusion Kalman filters.[1] Main idea is the use of mathematical methods in information integration and integrated technology tools of different sources of information to get high quality useful information.[2] Reference[3] was proposed based on the weighted least squares estimate, optimal linear unbiased estimation of unified optimal fusion rules;[4] is proposed based on linear mean square estimation of the optimal fusion rule, concluded that “using the measured information, the more the optimal fusion estimation of the amount of information is larger, the higher the precision of conclusions;[5] though laser multiplier method and matrix differential operators, respectively is proposed according to the weighted matrix, according to a scalar weighting and according to the scalar weighting of the components in the three kinds of linear minimum variance information fusion estimation criterion; Literature[6] of multisensor state of discrete linear time-delay stochastic system, put forward the augmented distributed
high real-time requirements of tracking system is not practical. In order to solve problems in the EKF, aiming to the Julier and Uhlmann made suitable for nonlinear systems with no trace of Kalman (Unscented Kalman Filter, UKF).[8] UKF was a group of Sigma points through deterministic sampling, so that they can get more observation assumes that increased the mean and covariance estimation accuracy of the system state. UKF without linear approximation of nonlinear systems, thus avoiding the linearization errors, and reduce the amount of intermediate computations. Because the UKF filter parameters are not sensitive, robust, high reliability.[9] Compare with EKF, UKF filter computation smaller estimates more accurate. For nonlinear systems, the UKF filter method based on information fusion filtering algorithm to further improve estimation accuracy.
