Parametric Point2Point2Range #398
Conversation
| function (cfo::CalcFactor{<:Point2Point2Range})(rho, xi, lm) | ||
| # Basically `EuclidDistance` | ||
| return [rho[1] - norm(lm .- xi)] | ||
| end |
There was a problem hiding this comment.
right, understand now. Best solution would be to drop the additional theta variable, but we should look at how best to do together with the Manifolds.jl upgrade -- lots of overlap.
There was a problem hiding this comment.
I never liked the additional theta thing. Perhaps the solution is just to skip immediately through to JuliaRobotics/IncrementalInference.jl#1051
and then we can drop the additional theta value since the inflated (high entropy) proposal will give the necessary spread.
|
Hi @Affie , with IIF v0.21 tagged tests starting to pass here again although "This and Masters" is failing in parametric. Perhaps just check that out when you get a chance please. |
|
Point2Point2Range test is failing because its not converging correctly every time, it depends on the initial values. It seems like such a small example that I’m sure fine tuning the solver parameters can make it better. |
|
Would adding a small perturbation similar to Roots fix the sporadic problem? |
|
We can now consolidate and remove extra |
|
Combined with #416 |
cc @dehann
PR for discussing what to do with factors like Point2Point2Range.
See
EuclidDistancelike factor and test for batch parametric solve