Issue with differential constraint from partial factors

[Affie]

Differential factors should be partial if source is only partial:

For example, if a child clique has a Pose2Point2Bearing at l1 | x1,x2 then the message can only be Pose2Pose2Bearing ie. partial on θ

Same for prior factors with partials.

EDIT: Needed for parametric tree solve.

[Affie]

This subfg from a range only example.

Results in these 2 factors after deconvolution.

x0 and x1 are connected by 2 range only factors and the differential factor cannot represent x1 fully.

[Affie]

For now useMsgLikelihoods=false should be used for better performance

useMsgLikelihoods=true

useMsgLikelihoods=false

[Affie]

Example plots above were adapted from the ranges examples from https://github.com/JuliaRobotics/RoME.jl/blob/master/examples/RangesExample.jl

EDIT: I just moved :l100 and :l3 for a clearer picture and used tighter covariances:

• GTp[:l100] = [-20.0;0]
• GTl[:l3] = [-30.0;20]

[dehann]

thanks, also linking some documentation on this that example for reference: https://juliarobotics.org/Caesar.jl/latest/examples/basic_slamedonut/

[dehann]

have not forgotten, just been a bit busy sorry. Will get to this soon!