[Newton] Fix find joints / body masks

[AntoineRichard]

Description

Fixes a bug in the mask creation when only a partial list of joints / body were passed.

Fixes #4439

Type of change

• Bug fix (non-breaking change which fixes an issue)

Checklist

• I have read and understood the contribution guidelines
• I have run the pre-commit checks with ./isaaclab.sh --format
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• I have updated the changelog and the corresponding version in the extension's config/extension.toml file
• I have added my name to the CONTRIBUTORS.md or my name already exists there

[greptile-apps]

Greptile Overview

Greptile Summary

Fixed critical indexing bug in find_joints and find_bodies functions when using subset parameters. The bug caused local indices (relative to the subset) to be returned instead of global indices (relative to the full list), resulting in incorrect mask values.

Core Fix:

• Modified find_joints and find_bodies in source/isaaclab_newton/isaaclab_newton/assets/utils/shared.py to map matched names back to global indices when a subset is provided
• Added body_subset parameter to find_bodies for consistency with find_joints
• Zero overhead when subset is not used (backwards compatible)

Additional Changes:
This PR also includes substantial refactoring and new features beyond the stated bug fix:

• New WrenchComposer utility class for force/torque composition
• Extensive test infrastructure additions (mock utilities, benchmarks, integration tests)
• Refactoring of RigidObjectData with new properties (is_primed, body_mass, body_inertia)
• Various improvements to articulation and rigid object implementations

Test Coverage:

• Comprehensive unit tests with 8+ test cases covering edge cases (non-contiguous subsets, partial matches, order preservation)
• Integration tests validating the fix works end-to-end with actual articulation objects
• Tests confirm fix for issue [Bug Report] find_joints in isaaclab_newton returns incorrect mask/indices when joint_subset is provided #4439

Confidence Score: 4/5

• This PR is safe to merge with minor review recommended for scope beyond the stated bug fix
• The core bug fix in shared.py is clean, well-tested, and directly addresses issue [Bug Report] find_joints in isaaclab_newton returns incorrect mask/indices when joint_subset is provided #4439. However, the PR bundles significant additional work (new WrenchComposer, extensive refactoring, test infrastructure) that wasn't mentioned in the PR description. The fix itself is solid with excellent test coverage, but the large scope increases merge risk.
• Pay attention to source/isaaclab_newton/isaaclab_newton/assets/rigid_object/rigid_object_data.py which has extensive refactoring including new properties and state management changes that warrant careful review

Important Files Changed

Filename	Overview
source/isaaclab_newton/isaaclab_newton/assets/utils/shared.py	Core bug fix: corrects find_joints/find_bodies to return global indices when subset is provided, fixing issue #4439
source/isaaclab_newton/test/assets/utils/test_shared.py	Comprehensive tests for the bug fix with excellent edge case coverage including non-contiguous subsets and order preservation
source/isaaclab_newton/isaaclab_newton/assets/articulation/articulation.py	Updates find_bodies signature to add body_subset parameter for consistency with find_joints, plus various refactoring changes
source/isaaclab_newton/isaaclab_newton/assets/rigid_object/rigid_object.py	Updates find_bodies signature to add body_subset parameter, plus extensive refactoring and improvements
source/isaaclab_newton/test/assets/articulation/test_articulation.py	Adds integration tests for find_bodies and find_joints with subset parameters to validate the fix works end-to-end

Sequence Diagram

sequenceDiagram
    participant User
    participant Articulation
    participant find_joints
    participant resolve_matching_names
    
    User->>Articulation: find_joints(".*", joint_subset=["front_left_leg", "front_left_foot", "left_back_leg", "left_back_foot"])
    Articulation->>find_joints: Call with joint_names=[all 8 joints], subset=[4 joints]
    
    Note over find_joints: search_target = joint_subset (4 joints)
    
    find_joints->>resolve_matching_names: Match ".*" in subset
    resolve_matching_names-->>find_joints: indices=[0,1,2,3] (local to subset), names=[4 matched joints]
    
    Note over find_joints: Bug was here - returned local indices<br/>Fix: Map names back to global indices
    
    find_joints->>find_joints: indices = [joint_names.index(n) for n in names]
    Note over find_joints: New indices=[0,1,4,5] (global positions)
    
    find_joints->>find_joints: Create mask with len(joint_names)<br/>Set mask[0,1,4,5] = True
    
    find_joints-->>Articulation: Return (mask, names, [0,1,4,5])
    Articulation-->>User: Global indices [0,1,4,5] ✓

Loading

[greptile-apps]

_{No files reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}

[kellyguo11]

Could we try to capture some of the other changes in the PR description as well? or maybe we can try splitting up the PR a bit. github seems to be refusing to show more than 1 file at a time because of the PR size.