State representation

[nahueespinosa]

Normally, MCL implementations (https://github.com/ekumenlabs/RnD/discussions/126#discussioncomment-3801233) tend to use ad-hoc structs to represent state. See:

• https://github.com/ros-planning/navigation2/blob/main/nav2_amcl/include/nav2_amcl/pf/pf_vector.hpp
• https://github.com/VorpalBlade/quickmcl/blob/master/include/quickmcl/pose_2d.h
• https://gitsvn-nt.oru.se/software/ndt_tools_public/-/blob/master/ndt_localization/include/ndt_localization/particle.hpp

Instead, let's select a library and use a fully featured vector representation for the state of the particles. Some available options are:

• https://github.com/strasdat/Sophus
• https://bitbucket.org/jmangelson/lie
• https://github.com/artivis/manif
• https://github.com/xtensor-stack/xtensor ?

A key feature for the chosen library would be to support Special Euclidean group SE(n) and Special Orthogonal group SO(n) in two and three dimensions.

A good definition of done for this ticket would be to implement a basic motion model using the chosen library and add it to the repository.

[hidmic]

It seems that Cami won't be tackling this any time soon.

[nahueespinosa]

From the three Lie theory libraries listed here, sophus is the only one already available in rosdep, so it feels like the natural choice.

[nahueespinosa]

Closed by #27.