Improve the implementation of euler_xyz_from_quat

[ShaoshuSu]

Description

Previously, euler_xyz_from_quat returned Euler angles in the [0, 2π) range, which is uncommon in robotics applications. As a result, several users implemented their own workarounds to adjust the angle range to (−π, π]. Examples include:

• Isaac-RL-Two-wheel-Legged-Bot: Custom unwrapping implemented in rewards.py#L17.
• lorenzo-bianchi/IsaacLab: Manual angle shifting in quadcopter_env_v1.py#L219.
• kscalelabs/klab: Custom angle normalization in imu.py#L17.

To address this issue, we have updated the default angle range from [0, 2π) to (−π, π] by removing the % (2 * torch.pi) operation at the return statement, since the atan2 function naturally outputs angles within the (−π, π] range.

We also introduced a new parameter, wrap_to_2pi: bool = False. Setting this parameter to True will maintain the previous behavior:

• Default (wrap_to_2pi=False): Angles returned in the range (−π, π].
• Optional (wrap_to_2pi=True): Angles wrapped in the original [0, 2π) range.

Additionally, multiple test samples have been added to evaluate and ensure performance and accuracy across different scenarios.

What’s Changed

• Default behavior updated to (−π, π] (non-breaking).
• New argument wrap_to_2pi for optional wrapping.
• Unit tests added for both wrapped and unwrapped outputs.

---

Fixes https://github.com/isaac-sim/IsaacLab/issues/2364

Type of Change

• Enhancement (non-breaking addition of functionality)

---

Checklist

• [x] Ran ./isaaclab.sh --format and all pre-commit checks pass
• [x] Updated docstrings to describe wrap_to_2pi parameter
• [x] Added tests covering both wrapped and unwrapped outputs
• [x] No new warnings introduced
• [ ] Updated changelog and bumped version in config/extension.toml
• [x] Confirmed my name is listed in CONTRIBUTORS.md

[ShaoshuSu]

No problem. Looking forward to these test upgrades! : )

[jtigue-bdai]

@ShaoshuSu we have landed the pytest infrastructure change. Please update your implementation when convenient.

[ShaoshuSu]

Hi @jtigue-bdai, I've completed the requested modifications. Please let me know if you have any further questions or suggestions.

By the way, I'm not sure whether the default value for wrap_to_2pi should be True (the previous default) or False (introduced in this PR). I know we often set wrap_to_2pi=False and expect the output range (-π, π]. However, are there specific scenarios where wrap_to_2pi=True would be preferred?

I'd appreciate your advice on this.

Thanks!