[maxkra15]

Summary

Reusable Newton fixes extracted from #7185. No environment, robot asset, example or video recorder is included.

• Include active cloth bending constraints in VBD coloring.
• Expose structural linear/angular joint stiffness independently of actuator gains, preserving native 1e5 defaults.
• Expose friction_epsilon, preserving 1e-2 m/s.
• Publish generalized state after standalone VBD physics ticks inside the CUDA graph. update_joint_state=False opts out; coupled and externally integrated rigid bodies retain their state ownership.
• Resolve configured nested USD articulation roots when constructing a Newton IK action prototype.
• Document structural versus actuator stiffness and NewtonCfg collision scheduling at their proper owners.

Related: #6733. Prerequisite for the refreshed #7185 example; independent of #8107. Migration and new runtime dependencies: none.

Type of change

• Bug fix
• Non-breaking configuration additions
• Documentation update

Validation

Merged upstream develop at dd2f470fcad5873980e9bc4acffb043238aca41c. Validated source committed as f2a6caef2e342d511c2b15677f24fe862fcfef66 on an NVIDIA L40, pinned Newton 1.7.0.dev0 (5ea82eef6f7911388025a8df6ccdb172e2fed286) and Warp 1.17.0. Each file ran separately with uv run --frozen --no-sync python -m pytest -q:

Test selection	Passed
source/isaaclab_newton/test/physics/test_vbd_core.py	9
source/isaaclab_newton/test/assets/test_articulation.py	38
source/isaaclab_newton/test/assets/test_rigid_object.py -k vbd_root	3
source/isaaclab_newton/test/controllers/test_newton_ik_solver.py	5
source/isaaclab_contrib/test/coupling/test_coupler_runtime.py	4
source/isaaclab_contrib/test/coupling/test_coupler.py	73

132 passed, zero failed or skipped. The new real-asset nested-root action test checks an independently derived 0.4-radian joint target; removing the root suffix fails at native articulation selection. Existing upstream deprecation/fixture warnings remain.

ISAACLAB_CHANGELOG_BASE_REF=origin/develop UV_FROZEN=1 uv run --no-sync isaaclab -f passed, including the changelog gate. A fresh warning-as-error documentation build passed using python -m sphinx -b html -W --keep-going -E -a -j auto docs docs/_build/support-review-guide-final.

The retained tests distinguish native defaults/configuration, analytic stiffness equilibrium, public state ownership and coupled/native parity. Newton owns offset/moving-parent joint reconstruction; no duplicate kinematics fixture or tablecloth rollout is added here.

Publication overhead was measured on the existing headless block-and-tackle demo on the same L40: four separate processes in off/on/on/off order, each with 40 warm-up frames and five 120-frame batches. The average process medians were 18.2855 ms/frame off and 18.3147 ms/frame on (difference 0.0292 ms, 0.16%). All runs captured a graph. This is a small-workload observation, not a significant difference or general throughput guarantee; GPU 1 was concurrently used for Kit rendering during later runs. The comparison changes only update_joint_state and synchronizes between batches, not per frame.

New-head CI is requested separately.

Checklist

• Read the contribution guide and followed package conventions.
• Ran formatting/lint, focused tests and the test-audit gate.
• Updated documentation and per-package fragments/minor marker.
• Existing entry in CONTRIBUTORS.md.

[maxkra15]

run-ci

[greptile-apps]

Confidence Score: 4/5

[High risk] Modifies VBD physics solver behavior and joint state synchronization.

Fix coupled managers' state-ownership policy before merging.

Findings

1.  Coupled solvers overwrite owned state ▶
2.  Articulated joint motion goes untested ▶

Summary

This PR adds cloth bending to VBD coloring, exposes structural joint stiffness, publishes joint state after body integration, and uses configured nested roots for Newton IK.

• Coupled managers inherit publication without applying their state-ownership settings.
• The publication regression covers a free root, but not joints between moving bodies.

Diagram

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[Physics tick] --> B[Run solver substeps]
    B --> C[Final state_0]
    C --> D{_update_joint_state?}
    D -->|Yes| E[eval_ik writes joint_q and joint_qd]
    D -->|No| F[Keep solver-owned joint state]
    G[Plain VBD build] --> H[Set flag from configuration]
    H --> D
    I[Coupled manager build] --> J[Inherit flag without setting ownership policy]
    J --> D

Loading

_{Reviews (1) · Last reviewed commit: "Fix VBD articulation state and cloth ben..." · Reviewed by Greptile}

[isaaclab-review-bot]

Isaac Lab Review Bot

Needs attention: Style consistency (2). Other reviewed categories look good.

• PR description: Description matches the diff and reports workload, warmup, and timings; add the benchmark command, hardware, driver, environment count, revisions, and pinned Newton version for reproducibility.
• Design and architecture: Architecture looks good.
• Scope and over-engineering: No over-engineering issues.
• API: API changes look good.
• Compatibility and deprecation: No breaking changes identified.
• Implementation: Implementation looks good.
• Style consistency: The added own-scene VBD test makes the module-level scene-ownership description stale; update that description to include this case.
• Test quality: Test quality looks good.

Minor fixes needed. Posted 2 inline findings.

Automated review; human maintainers own approval decisions.

[mmichelis]

Thanks for adding the option for eval_ik!

[maxkra15]

run-ci

[maxkra15]

run-ci