Dehann Fourie
Dehann Fourie
After completing #459 rework, we should improve the LikelihoodMessage struct
## TODOs - [x] Improve test coverage in IIF, https://codecov.io/gh/JuliaRobotics/IncrementalInference.jl/src/master/src/TetherUtils.jl - [ ] Spruce up related docs in Caesar, https://juliarobotics.org/Caesar.jl/latest/examples/deadreckontether/ ## RoME already has Pose2 test https://github.com/JuliaRobotics/RoME.jl/blob/master/test/testDeadReckoningTether.jl ## Notes important...
```julia hists = IIF.fetchCliqHistoryAll!(smtasks) IIF.printCSMHistorySequential(hists, 4=>1:10) IIF.printCSMHistoryLogical(hists, [4;7]=>1:10) ``` Using such a filter on `Logical` is not working. Need to accept both sides of the `Pair` any combination of `Int,...
Before some computations, make sure CSM cliqSubFg does not contain reserved variable or factor tags like :LIKELIHOODMESSAGE or UPWARD_DIFFERENTIAL EDIT: see update #1048
Post #913 and #923 , the old way of deciding which variables to `iterVar` or solve `direct` inside a clique during upsolve is likely no longer correct. Should use `csmc.cliqSubFg`...
Perhaps generalize to return full measurement tuple and not just meas[1] values. Part of larger factor standardization effort.
Parent clique: - `x34, x9, x54, x26, x46, x14, x4` with 5 children: - `x38 | x34, x4, x26, x54, x46, x14` - `x30 | x26, x34` - `x50 |...
Want to drop the dependency entirely. Currently only giving the user the option to use ccolamd, but its not default.
Some factors may have user defined configuration files, that although saved in the jld's, change over time and are no longer compatible with the new computational types. Usual options include...