search

RobotLocomotion / drake

Model-based design and verification for robotics.
https://drake.mit.edu
Other
3.29k stars 1.26k forks source link

Use terms inboard/outboard instead of parent/child in internal MBT code #12789

Open amcastro-tri opened 4 years ago

amcastro-tri commented 4 years ago

Per review comment here, the parent/child terminology might be somewhat fuzzy inside a context that already uses that terminology (for instance MBT parent of a body, or parent/child bodies in an SDF joint). Therefore it'd be a good idea to refactor this internal MBT code to use the inboard/outboard terminology, which is widely accepted in the community.

This could be a great first PR for Joe!

mitiguy commented 4 years ago

The inboard/outboard terminology and related issues (#12789 and #13040) arose in PR# 13961. it would be helpful to first agree/converge on frame names for documentation and notation.
For example, we currently use Parent body P with frame F and Child body B with frame M. Perhaps this should be Parent body P with frame Jp and Child body C with frame Jc. For topology, perhaps we could use different words such as A/B or inboard/outboard or upstream/downstream or ...

amcastro-tri commented 4 years ago

I've been consistently using P/F with B/M at the "mobilizer" level (internal tree model of our world) and P/Jp and B/Jc at the "joint" level (physical representation of our world). If somewhere in Drake you find an inconsistency, it should be fixed. But I believe we did like this in the past, given that mobilizers will eventually be reversible in order to enable the modeling of arbitrary loop kinematics.

PS. I removed comments from #13040 since there is no need to have the same discussion sprinkled around. Besides, it belongs here.

sherm1 commented 4 years ago

I'll assign this one to @mitiguy for now since he's currently working on joint terminology docs.