Upstream PR: rlabbe/filterpy#301
Issue Type: Cherry-pick failed: fatal: bad object b3b387b8edc5d1d205206d5e432a6b20d04c404a
Problem
Error: Cherry-pick failed: fatal: bad object b3b387b8edc5d1d205206d5e432a6b20d04c404a
Original Description
- The original EKF assumes a linear, time invariant (LTI) transition matrix. That is, in general, not a requirement for an EKF. Since the measurement Jacobian can be overridden, the transition matrix should also be able to be computed at each step. This PR adds an override to enable that capability.
- The original install method was via setup.py only. This doesn't work with newer build systems that are using TOML files, as the dependencies don't seem to be handled correctly. This PR updates the setup.py file to be a TOML file instead to correctly handle dependencies.
- The README docs mentioned that Python 2.7 and 3.4 support was supposed to be dropped around 2018. The setup.py file doesn't show this. As part of updating to the TOML file, the classifiers were also updated to be Python 3.7.
- As part of the TOML file, I did add in flake8 and autopep definitions. They are not being used to lint currently, but provide the basis for linting if desired.
Upstream PR: rlabbe/filterpy#301
Issue Type: Cherry-pick failed: fatal: bad object b3b387b8edc5d1d205206d5e432a6b20d04c404a
Problem
Error: Cherry-pick failed: fatal: bad object b3b387b8edc5d1d205206d5e432a6b20d04c404a
Original Description