filterpy
filterpy copied to clipboard
rlabbe
Python Kalman filtering and optimal estimation library. Implements Kalman filter, particle filter, Extended Kalman filter, Unscented Kalman filter, g-h (alpha-beta), least squares, H Infinity, smoothe...
The code uses self.S = P_inv + dot(H_T, R_inv).dot(H) which is consistent with Dan Simon's book, but I am dubious. In general the shape of covariance in state space and...
It uses the standard NumPy formatting convention of x : numpy.array(dim_x, 1) State estimate vector but displays with type intermixed with description: x | (numpy.array(dim_x, 1)) state estimate vector
This makes HInfinity filter unable to use in IMM Estimator.
The EKF does not implement batch_filter or rts. There are probably similar instances of differences between the classes.
I haven't decided on the version to support. 3.5 at the minimum, though 3.6 is very appealing to me due to f-strings. At a minimum, this change would remove all...
Strictly speaking it is not needed, since x, P are computed in update() from the bank of filters, but users may expect an updated prior after predict() is called.
I just ran the example in the UKF.py docstring, and the log-likelihood are in the range of -20000 or so, Either the computation is incorrect or the problem formulation is...
See https://github.com/rlabbe/Kalman-and-Bayesian-Filters-in-Python/issues/188, which refers to usage in the book as well as this library. There are several functions in stats.py that compute things that are implemented in NumPy or SciPy....