Fix star terrain crash on NumPy 2

[j3soon]

Description

star_terrain computed the bar lengths with np.math.cos and np.math.sin. np.math was an alias for the standard library math module; it was deprecated in NumPy 1.25 and removed in NumPy 2.0. Since Isaac Lab 3.0 ships NumPy 2, any terrain configuration containing the star sub-terrain raised AttributeError: module 'numpy' has no attribute 'math', making MeshStarTerrainCfg unusable.

Use the standard library math module instead. The alias was the same module object, so the generated geometry is unchanged.

The regression went unnoticed because ROUGH_TERRAINS_CFG, the only config covered by the terrain generator tests, has no star sub-terrain. Add a test that generates a star-only terrain, with a bar count that exercises all three branches of the bar-length computation.

Note: The testcase is written by Claude. I've confirmed the code fix myself.

Fixes #6909

Type of change

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

Release backport

• Backport this pull request to the active release branch after it merges into develop

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 added a changelog fragment under source/<pkg>/changelog.d/ for every touched package (do not edit CHANGELOG.rst or bump extension.toml — CI handles that)
• I have added my name to the CONTRIBUTORS.md or my name already exists there

[greptile-apps]

Greptile Summary

This PR restores star-terrain generation on NumPy 2 by replacing the removed np.math alias with Python’s standard math module.

• Updates all three trigonometric bar-length branches without changing their geometry.
• Adds a star-only terrain generation test that exercises every branch and validates mesh dimensions and origin shape.
• Documents the NumPy 2 compatibility fix in a changelog fragment.

Confidence Score: 5/5

The PR appears safe to merge with no actionable defects identified.

The standard-library functions are equivalent to the removed NumPy aliases, and the new regression configuration and assertions align with the terrain-generation pipeline.

Important Files Changed

Filename	Overview
source/isaaclab/isaaclab/terrains/trimesh/mesh_terrains.py	Replaces removed NumPy aliases with equivalent standard-library trigonometric functions, preserving the existing bar-length calculations.
source/isaaclab/test/terrains/test_terrain_generator.py	Adds valid star-terrain regression coverage whose five bars exercise all three changed branches and whose assertions match the generator’s 1x1 output.
source/isaaclab/changelog.d/j3soon-star-terrain-numpy2.rst	Accurately documents the NumPy 2 failure and compatibility fix.

_{Reviews (1): Last reviewed commit: "Fix star terrain crash on NumPy 2" | Re-trigger Greptile}

[isaaclab-review-bot]

Isaac Lab Review Bot

The patch replaces the NumPy 2-incompatible np.math access in star_terrain with the equivalent standard-library math functions, adds focused star-terrain generator coverage, and includes the required patch changelog fragment.

• Design and architecture: The change remains localized to the existing bar-length calculation and preserves terrain configuration, registration, and generator ownership. The regression test appropriately exercises the public TerrainGenerator path with MeshStarTerrainCfg.
• API: No public API signatures, symbols, or return contracts change. Because np.math previously exposed the standard-library math module, switching to math.cos and math.sin preserves the intended scalar computation while restoring NumPy 2 compatibility.
• Implementation: The num_bars=5 test configuration reaches all three yaw branches and validates successful generation through mesh dimensions and terrain-origin shape. NumPy remains required elsewhere in the implementation. The changelog fragment is correctly categorized and formatted. Pre-commit was not reported as run, so CI formatting and lint verification remains outstanding.

No blocking issues. No inline issue met the actionable-evidence threshold; the assessment above records the review feedback.

Automated review; human maintainers own approval decisions.

[AntoineRichard]

Thanks for fixing this and adding focused regression coverage! I verified that the new test passes on this commit and fails on the base with the expected NumPy 2 error, and the full pre-commit suite passes. Approved.

[AntoineRichard]

@StafaH @ooctipus can we get one more review to get it merged in.

[AntoineRichard]

run-ci

[isaaclab-bot]

Backported to release/3.0.0 as c7fd163.