removing world context

[danbarla]

changes included in first commit:

1. Link class refactored to hold one member : bMcom - link com in base frame at rest. wTl and lTcom removed. Constructor adjusted.
2. sdf reading functions adjusted. joints are defined in link frames (not COM) so we need to read the bTl in some places.
3. tests adjusted.

changes in second commit:

1. only change of naming in few files from wTj to bTj.

still missing:

1. need to adjust interface file (.i)
2. check python wrapper and tests
3. wTcom function still exists in Link class.

[dellaert]

@gchenfc and @yetongumich please fix CI to include jumping robot and cable robot ASAP? we will merge and break your code if CI says that’s OK.

[dellaert]

I think you are still working on this so just re-request a review when you need it.

[dellaert]

@danbarla Please (a) merge in master (b) let me know in reply what things are still missing from this PR (if any).

[danbarla]

@dellaert left to do:

1. examples/example_quadruped_mp/main.cpp : wTb is still there in many places.
2. tests/testInitializeSolutionUtils.cpp : wTb and wTl still widely used there
3. add unit tests for the bMcom of the "body" link to be identity
4. interface file, python wrapper and python tests.

do I need to open a new PR that merges my branch to master?

[dellaert]

do I need to open a new PR that merges my branch to master?

@danbarla you need to merge master into this branch, not the other way around. That way you get the latest updates from @varunagrawal and @gchenfc that will run the python unit tests in CI. To do this, open a terminal, make sure you are at the latest version of the branch, and do

git status # to make sure you’re in the branch
git pull
git merge master

[dellaert]

PS You might get conflicts, but probably not. I actually rarely use the command line, instead I use smartgit (free for non-commercial use) which makes it super easy to resolve conflicts.

[danbarla]

so I merged master to my branch. testContactPointFactor.cpp doesn't pass now (it uses wtl(), I think its a new test) I am also having some trouble with the python tests.

[dellaert]

Maybe we should try to meet? Send me an email

[varunagrawal]

@dellaert did we ever resolve the question about computing contact points if we don't have the wTl/bTl methods?

Another use case I've encountered is when using other people's code. E.g. when I use Pybullet, everything is in wTl and not wTcom so this requires me to manually add the transforms for lTcom. I imagine this will be the case for multiple scenarios and would affect interoperability with other libraries.

[dellaert]

@varunagrawal We will create an issue to bring back bTl as bMlink in a separate PR.