Nonlinear Filtering

Continuous Time H-Infinity Filter with Asymptotic Convergence

“…This process requires assumptions on the statistics of the process noise and measurement noise (Raol, 2017). Often such information is not available.…”

Fuzzy Logic Augmented Observer and Stability Analysis for Image - Centroid Tracking

“…The value of R (or 'Rr') appearing in (5) is used as a tuning parameter. In order to implement the algorithm, one has to solve the matrix ORD equation (6), and for this one can use the following transformation [9] a=P(t) d (24) to obtain the differential equations, appropriately from (4), as follows (25) and (26) are solved by using the transition matrix method [9], then using (24) one gets P(t). Model of FLIR (forward looking infrared sensor) for generation of synthetic image is considered.…”

“…In the case of the observer of (3), R in (5) is some positive definite matrix, and is regarded as the weighting matrix. In (3), f m is an FMF that operates on input residuals r(t), (in turn on e(t)) and obtains the output that is used in the observer structure; this provides a nonlinear function of residuals effect [9,Chapter 9], because any FMF is inherently nonlinear. In (5) P(t) is obtained as the solution of the observer Riccati differential (ORD) equation (6), Q is considered as some weighting matrix, whereas in KF/EKF theory it is an intensity or a covariance matrix of the process noise.…”