Use //link/inertial/density for auto-inertials

[scpeters]

🎉 New feature

Closes #1329. cc @jasmeet0915

Summary

This adds support for specifying density and auto_inertia_params in the //link/inertial/density and //link/inertial/auto_inertia_params elements respectively, which will apply to all collisions within a link. The following APIs are added:

• static double Collision::DensityDefault(): accessor for default density value
• overload of Collision::CalculateInertial with a std::optional<double> input to represent the value from //link/inertial/density if it has been set and an sdf::ElementPtr to the <auto_inertia_params> element or nullptr if it has not been set
• double Link::Density() const
• void Link::SetDensity(double)
• sdf::ElementPtr AutoInertiaParams() const
• void SetAutoInertiaParams(const sdf::ElementPtr)

The density precedence for Collisions is now:

1. Density explicitly set in //collision/density or by Collision::SetDensity.
2. Density explicitly set in //link/inertial/density or by Link::SetDensity.
3. Default density in Collision::DensityDefault().

The following table summarizes this behavior:

Density to use	Collision density set	Collision density not set
Link density set	Collision density	Link density
Link density not set	Collision density	Default density

The precedence rules for AutoInertiaParams match the rules for density.

Test it

A test has has been added in UNIT_Link_TEST with the 4 combinations of link / collision density being set or not from the table above. Also the test/sdf/inertial_stats_auto.sdf file, which is tested with gz sdf --inertial-stats in UNIT_gz_TEST, is modified so that a //collision/density value is converted to //link/inertial/density, and the test still passes.

Checklist

• Signed all commits for DCO
• Added tests
• Added example and/or tutorial
• Updated documentation (as needed)
• Updated migration guide (as needed)
• Consider updating Python bindings (if the library has them)
• codecheck passed (See contributing)
• All tests passed (See test coverage)
• While waiting for a review on your PR, please help review another open pull request to support the maintainers

Note to maintainers: Remember to use Squash-Merge and edit the commit message to match the pull request summary while retaining Signed-off-by messages.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (c3960a5) 92.40% compared to head (5bd4dc2) 92.42%.
Report is 6 commits behind head on sdf14.

Additional details and impacted files

@@            Coverage Diff             @@
##            sdf14    #1335      +/-   ##
==========================================
+ Coverage   92.40%   92.42%   +0.01%     
==========================================
  Files         134      134              
  Lines       17674    17707      +33     
==========================================
+ Hits        16332    16365      +33     
  Misses       1342     1342              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[azeey]

Per #1330 (comment), and #1330 (comment) could we pass along the //link/inertial/auto_inertia_params in the new member function and avoid needing yet another overload?

[scpeters]

Per #1330 (comment), and #1330 (comment) could we pass along the //link/inertial/auto_inertia_params in the new member function and avoid needing yet another overload?

good call, thanks for the reminder

I've started work in scpeters/inertial_density...scpeters/inertial_density_auto_params, but I didn't write the test yet, since I think it will need a mock Geometry class in order to confirm which ElementPtr is passed to CalculateInertial.

[azeey]

Per #1330 (comment), and #1330 (comment) could we pass along the //link/inertial/auto_inertia_params in the new member function and avoid needing yet another overload?

good call, thanks for the reminder

I've started work in scpeters/inertial_density...scpeters/inertial_density_auto_params, but I didn't write the test yet, since I think it will need a mock Geometry class in order to confirm which ElementPtr is passed to CalculateInertial.

Instead of a mock Geometry class, would it be easier to provide a mesh inertia calculator callback in the test that will receive the autoInertiaParams and check to see if it's the one from the link? i.e,

sdformat/src/Geometry.cc

Line 341 in 11205c9

this->dataPtr->mesh->CalculateInertial(_errors, _density,

will eventually call the callback provided by the test with the autoInertiaParams passed to Geometry.

[scpeters]

Instead of a mock Geometry class, would it be easier to provide a mesh inertia calculator callback in the test that will receive the autoInertiaParams and check to see if it's the one from the link? i.e,

sdformat/src/Geometry.cc

Line 341 in 11205c9

this->dataPtr->mesh->CalculateInertial(_errors, _density,

will eventually call the callback provided by the test with the autoInertiaParams passed to Geometry.

I adapted the lambda callback from Geometry_TEST.cc for an auto_inertia_params test in 4eae4b8 and merged in the other changes from that branch

[scpeters]

I fixed a compiler warning and coverage; CI is happy now