Add mesh optimization attribute to <mesh>

[iche033]

🎉 New feature

Replaces #1380

Implemented proposal in #68

Summary

Adds a new optional attribute called simplification optimization to the <mesh> sdf element. This allows users specify a method for simplifying the mesh, e.g.

<collision name="my_collision">
  <geometry>
    <mesh optimization="convex_decomposition">
      <uri>path_to_complex_mesh</uri>
    </mesh>
  </geometry>
</collision>

Also updated the Mesh DOM class to include this new attribute.

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.

[iche033]

copying the comment from #1380 for visibility

@scpeters:

I'm trying to decide if simplification is the right name for this attribute.

Alternatives (which might not be better):

//mesh/@transformation
//mesh/@convexity
with candidate values:

use_convex_hull
use_convex_decomposition

@iche033:

I think transformation is generic and would be another good candidate. So we don't limit this to just creating convex meshes

thoughts? @azeey

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.44%. Comparing base (b04828d) to head (1b0736e).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1382      +/-   ##
==========================================
+ Coverage   92.42%   92.44%   +0.02%     
==========================================
  Files         134      134              
  Lines       17751    17812      +61     
==========================================
+ Hits        16406    16467      +61     
  Misses       1345     1345              

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

[azeey]

copying the comment from #1380 for visibility

@scpeters:

I'm trying to decide if simplification is the right name for this attribute.
Alternatives (which might not be better):
//mesh/@transformation
//mesh/@convexity
with candidate values:
use_convex_hull
use_convex_decomposition

@iche033:

I think transformation is generic and would be another good candidate. So we don't limit this to just creating convex meshes

thoughts? @azeey

transformation seems too generic. Would we including "scaling" for example?

Another option is "optimization"

[iche033]

Another option is "optimization"

"optimization" sounds good.

[iche033]

Another option is "optimization"

"optimization" sounds good.

checking to see if @scpeters if you're ok with the attribute name?

with candidate values:
use_convex_hull
use_convex_decomposition

Do we need the use_ prefix? I think it would only be needed if there's a case in which we plan to use a different verb?

[scpeters]

Another option is "optimization"

"optimization" sounds good.

checking to see if @scpeters if you're ok with the attribute name?

//mesh/@optimization sounds great

with candidate values:
use_convex_hull
use_convex_decomposition

Do we need the use_ prefix? I think it would only be needed if there's a case in which we plan to use a different verb?

no, we don't need use_

[scpeters]

do we want to expose the maxConvexHulls and voxelResolution parameters from gazebosim/gz-common#585 in the spec? if not, then perhaps we could read them using the gz: xml namespace in gz-physics

[iche033]

do we want to expose the maxConvexHulls and voxelResolution parameters

I think the maxConvexHull param can be exposed as that's pretty generic. The voxel resolution on the other hand would only be applicable if the algorithm uses a voxel based approach for representing the mesh before decomposition. Another interesting library I found is CoACD that claims better results than V-HACD and uses 3D cutting plans for deciding how to split the mesh.

I think I could do some research first to see how we can generalize the params before adding them to the spec.

[iche033]

ok updated to replace simplification with optimization. 75ca10e

think the maxConvexHull param can be exposed as that's pretty generic.

Updated mesh sdf spec with a new optional <convex_decomposition> element that exposes the max convex hull param. We can add other params later once we decide on them. b399bcd

<mesh optimization="convex_decomposition">
  <convex_decomposition>
    <max_convex_hulls>16u</max_convex_hulls>
  </convex_decomposition>
  ...
</mesh>

This PR is ready for review again.

[scpeters]

just a few small suggestions to improve coverage

[iche033]

just a few small suggestions to improve coverage

thanks, applied suggested changes in f82be1f

[scpeters]

I pushed one more improvement to code coverage in 1b0736e

[scpeters]

thanks for the patience with all my feedback, looks good!

[scpeters]

FYI I opened an out-of-scope feature request to specify assert_convex for a mesh in #1384

[iche033]

thanks for the reviews!