Physics System populates link bounding boxes #2821

gabrielfpacheco · 2025-03-12T17:43:36Z

🎉 New feature

Follow-up of #2787

Summary

Computes the link bounding boxes for all link entities that have components::AxisAlignedBox present. The computation follows the same logic as the models bbox, but falls back to the SDF calculation introduced in #2787 in case the feature is not present for the selected physics plugin.

Test it

Run the following integration test:

./build/gz-sim9/bin/INTEGRATION_physics_system --gtest_filter=*LinkBoundingBox*

It will load bounding_boxes.sdf world (see below) and perform link AABB checks while the models free-fall until reaching the ground.

Note: after the previous PR is merged, this PR will rebase to gz-sim9 and be marked as ready to review

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.

Signed-off-by: Gabriel Pacheco <[email protected]>

* Following gazebosim#2353 (comment) * This allows to use the ECM as cache if bounding box doesn't have to be recomputed and is more suitable so that other systems can determine the AABB if necessary (e.g. physics system if checks are enabled) Signed-off-by: Gabriel Pacheco <[email protected]>

This could be helpful if users want to explicetly compute the AABB from the link SDF collision shapes. For instance, the Physics system can fallback and use this method in case the selected engine does not support AABB calculation. Signed-off-by: Gabriel Pacheco <[email protected]>

Signed-off-by: Gabriel Pacheco <[email protected]>

iche033

did a first pass review, overall approach looks good

iche033 · 2025-03-14T18:50:33Z

src/Link.cc

+
+//////////////////////////////////////////////////
+void Link::EnableBoundingBoxChecks(
+  gz::sim::EntityComponentManager & _ecm,


we can omit gz::sim:: or sim:: since there are using namespace directives above in the file.

Addressed in 72e537a

iche033 · 2025-03-14T18:52:07Z

include/gz/sim/Util.hh

+    /// \param _sdfMesh Mesh SDF DOM.
+    /// \return The AABB of the mesh in its local frame.
+    GZ_SIM_VISIBLE math::AxisAlignedBox meshAxisAlignedBox(
+      const sdf::Mesh _sdfMesh);


can this be made const &?

Oops, that was a typo that has been kept unintentionally, thanks for catching.
Addressed in 84e70e7

iche033 · 2025-03-14T18:52:26Z

src/Util.cc

@@ -989,6 +989,38 @@ const common::Mesh *optimizeMesh(const sdf::Mesh &_meshSdf,
  return optimizedMesh;
 }

+math::AxisAlignedBox meshAxisAlignedBox(sdf::Mesh _sdfMesh)


make this const &?

Oops.. Addressed in 84e70e7

iche033 · 2025-03-14T19:03:48Z

test/integration/physics_system.cc

+  EXPECT_GT(duckBbox.Max().Y(), 0);
+}
+
+


remove empty line

Amended 0595ea9 to address this

iche033 · 2025-03-14T19:05:08Z

src/Util.cc

@@ -989,6 +989,38 @@ const common::Mesh *optimizeMesh(const sdf::Mesh &_meshSdf,
  return optimizedMesh;
 }

+math::AxisAlignedBox meshAxisAlignedBox(sdf::Mesh _sdfMesh)


consider also returning std::optional to indicate that the axis aligned box can not be computed

Addressed in 84e70e7

iche033 · 2025-03-14T19:07:25Z

src/systems/physics/Physics.cc

+          gzdbg << "Attempting to get a bounding box, but the physics "
+                 << "engine doesn't support feature "
+                 << "[GetLinkBoundingBox]. Link bounding boxes will be "
+                 << "computed from their collision shapes." << std::endl;


Suggested change

<< "computed from their collision shapes." << std::endl;

<< "computed from their collision shapes based on their "

<< "geometry properties in SDF." << std::endl;

Amended 0595ea9 to address this.

Signed-off-by: Gabriel Pacheco <[email protected]>

* Also fix `_sdfMesh` to be `const &` Signed-off-by: Gabriel Pacheco <[email protected]>

Signed-off-by: Gabriel Pacheco <[email protected]>

gabrielfpacheco · 2025-03-25T18:09:47Z

@iche033, I believe I addressed all your comments. If you agree with the proposed changes, could you please approve #2787 before? I would like to rebase this PR after the previous one is merged so that it focuses only on the physics system changes.

iche033 · 2025-03-26T17:39:17Z

@iche033, I believe I addressed all your comments. If you agree with the proposed changes, could you please approve #2787 before? I would like to rebase this PR after the previous one is merged so that it focuses only on the physics system changes.

Currently there's a build error as it needs the new API from sdformat. So I'm waiting for gazebosim/sdformat#1547 to be merged first. We also need a new sdformat release afterwards in order to run CI in #2787 and make sure everything is green. We can only approve and merge a PR when the required CI checks are green.

gabrielfpacheco · 2025-03-28T21:16:32Z

So I'm waiting for gazebosim/sdformat#1547 to be merged first. We also need a new sdformat release afterwards in order to run CI in #2787 and make sure everything is green.

Yes, sure! I initially thought there might be a way to trigger the CI with dependent packages at specific branches, but then realized that the merge/release process is required, as you said.

Besides the dependency that prevents the proper CI checks, though, there are still some unresolved threads here from your first pass review, are those solved?

Regarding the sdformat PR, the CI is already green, and I believe all comments have been addressed. I've pinged the reviewers in gazebosim/sdformat#1547 (comment) recently, so I think we should be close to a merge.

gabrielfpacheco added 8 commits March 10, 2025 09:56

Add support to get link AABB from its collisions

61b2e1a

Signed-off-by: Gabriel Pacheco <[email protected]>

Add python bindings to new methods and corresponding tests

dca7e4a

Signed-off-by: Gabriel Pacheco <[email protected]>

Fix and improve new methods API documentation

4aa7981

Signed-off-by: Gabriel Pacheco <[email protected]>

Include missing used headers

2c02ec1

Signed-off-by: Gabriel Pacheco <[email protected]>

Add python bindings to new public method

2168cff

Signed-off-by: Gabriel Pacheco <[email protected]>

Fix mesh AABB calculation to take scale factor into account

38d9fe0

Signed-off-by: Gabriel Pacheco <[email protected]>

github-actions bot added the 🏛️ ionic Gazebo Ionic label Mar 12, 2025

gabrielfpacheco mentioned this pull request Mar 12, 2025

Add support to get link AABB from its collisions #2787

Open

9 tasks

iche033 reviewed Mar 14, 2025

View reviewed changes

iche033 added the needs upstream release Blocked by a release of an upstream library label Mar 14, 2025

gabrielfpacheco added 3 commits March 14, 2025 17:44

Remove unnecessary gz::sim and sim:: namespaces

72e537a

Signed-off-by: Gabriel Pacheco <[email protected]>

Return std::optional to indicate mesh AABB cannot be computed

84e70e7

* Also fix `_sdfMesh` to be `const &` Signed-off-by: Gabriel Pacheco <[email protected]>

Compute link bounding box on physics system

0595ea9

Signed-off-by: Gabriel Pacheco <[email protected]>

gabrielfpacheco force-pushed the gp-link_aabb_physics_system branch from 1fed07c to 0595ea9 Compare March 14, 2025 20:52

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Physics System populates link bounding boxes #2821

Physics System populates link bounding boxes #2821

gabrielfpacheco commented Mar 12, 2025 •

edited

Loading

iche033 left a comment •

edited

Loading

iche033 Mar 14, 2025

gabrielfpacheco Mar 14, 2025

iche033 Mar 14, 2025

gabrielfpacheco Mar 17, 2025 •

edited

Loading

iche033 Mar 14, 2025

gabrielfpacheco Mar 14, 2025

iche033 Mar 14, 2025

gabrielfpacheco Mar 14, 2025

iche033 Mar 14, 2025

gabrielfpacheco Mar 14, 2025

iche033 Mar 14, 2025

gabrielfpacheco Mar 14, 2025

gabrielfpacheco commented Mar 25, 2025

iche033 commented Mar 26, 2025

gabrielfpacheco commented Mar 28, 2025

	<< "computed from their collision shapes." << std::endl;
	<< "computed from their collision shapes based on their "
	<< "geometry properties in SDF." << std::endl;

Physics System populates link bounding boxes #2821

Are you sure you want to change the base?

Physics System populates link bounding boxes #2821

Conversation

gabrielfpacheco commented Mar 12, 2025 • edited Loading

🎉 New feature

Summary

Test it

Checklist

iche033 left a comment • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

gabrielfpacheco Mar 17, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

gabrielfpacheco commented Mar 25, 2025

iche033 commented Mar 26, 2025

gabrielfpacheco commented Mar 28, 2025

gabrielfpacheco commented Mar 12, 2025 •

edited

Loading

iche033 left a comment •

edited

Loading

gabrielfpacheco Mar 17, 2025 •

edited

Loading