Constraint Manifold - Githubissues

borglab / GTDynamics

Full kinodynamics constraints for arbitrary robot configurations with factor graphs.

BSD 2-Clause "Simplified" License

40 stars 10 forks source link

Constraint Manifold #354

Closed yetongumich closed 1 year ago

yetongumich commented 2 years ago

Add constraint manifold optimizer implementation as in ManifoldOptimzier files
ConnectedComponent and ConstraintManifold are used to represent variables connected by constraints
Kinematic and dynamic planning examples as two cpp files in scripts

varunagrawal commented 2 years ago

Any reason this hasn't been reviewed yet? If the speedup potential is promising, I need it for the legged state estimator.

yetongumich commented 2 years ago

Did some clean up by

Inherit ManifoldOptimizer from ConstrainedOptimizer
Create struct ManifoldOptProblem to represent the manifold optimization problem
Explicitly shows the problem transform from equality-constrained optimization to manifold optimization problem

yetongumich commented 2 years ago

@gchenfc @varunagrawal @danbarla Would you like to do a review of the updated PR and see if it is ready to merge in?

yetongumich commented 2 years ago

@gchenfc @varunagrawal Can either of you approve this PR so that we can merge it into the master?

varunagrawal commented 2 years ago

@yetongumich I'll review it by tonight.

yetongumich commented 1 year ago

@varunagrawal @ProfFan It seems a refactorization in gtsam eliminates boost, which makes GTDynamics fail. Do we want to address it in this PR or a different PR?

gchenfc commented 1 year ago

@dellaert mentioned in a private conversation that we should consider using release/4.2 temporarily (e.g. in CI) while we migrate.

I think it (migrating) should definitely be a separate PR since there's lots of unrelated code in GTDynamics that uses boost::optional etc. As for updating the CI to use release/4.2, I guess still a separate PR probably makes the most sense... (my personal opinion)

yetongumich commented 1 year ago

Sounds good! I think I'll just leave the PR as it is since it cannot pass the CI check.

varunagrawal commented 1 year ago

@yetongumich why not go ahead and make a PR against this branch with the updated CI? I already fixed some of the compile issues against 4.2.

varunagrawal commented 1 year ago

Approving since we can make changes along the way.

varunagrawal commented 1 year ago

@yetongumich I used this PR to remove additional boost stuff.