[Fix] Unbreak the add_new_robot tutorial

[hujc7]

Summary

scripts/tutorials/01_assets/add_new_robot.py crashes on startup on develop and
release/3.0.0, and aborts on the first step for any --num_envs greater than one.
Two lines in the tutorial; no library change.

Description

The tutorial constructed its scene configuration positionally:

scene_cfg = NewRobotsSceneCfg(args_cli.num_envs, env_spacing=2.0)

@configclass builds a dataclass, so field declaration order is the positional
__init__ signature. Commit 0a0dc1b ("Support Multi-Backend for Demo Scripts", #5938)
added class_type as the first field of InteractiveSceneCfg, so num_envs began
landing in the class_type slot:

TypeError: Missing values detected in object NewRobotsSceneCfg for the following fields:
  - num_envs

Fixes NVBug 6652234.

Changes

1. Pass num_envs by keyword. This is the form every other scene-configuration
   construction in the repository and all five documented examples already use.
   class_type stays first in InteractiveSceneCfg, consistent with AssetBaseCfg,
   ActuatorBaseCfg, ActionTermCfg, CommandTermCfg and RecorderTermCfg.
2. Size the wheel-velocity action templates to scene.num_envs. A second, independent
   defect meant --num_envs greater than one aborted on the first step with
   Shape mismatch: torch.Size([1, 2]) != (2, 2).

Not addressed here: both standalone-script jobs pass standalone-script-scope: "demos",
so select_script_scope filters every scripts/tutorials path out of the launch matrix.
Tutorials are discovered and marker-checked but never launched, which is why this survived
to SQA. Left as-is; the tutorials are slated for removal.

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
• No documentation change is required; the tutorial text already describes keyword construction
• My changes generate no new warnings
• No test added: the fix is a three-line correction to a tutorial script that CI does not launch
• A package changelog fragment is not required because no source package changed
• I have added my name to the CONTRIBUTORS.md or my name already exists there

[hujc7]

run-ci

[hujc7]

Believe the tutorial is getting removed? fixing anyway for now. @StafaH

[isaaclab-review-bot]

Isaac Lab Review Bot

The tutorial now constructs NewRobotsSceneCfg with an explicit num_envs keyword and sizes both Jetbot wheel-velocity targets for every configured environment. The changes directly address constructor-order sensitivity and multi-environment target-shape handling.

• Design and architecture: Using a keyword for num_envs removes the tutorial’s dependence on inherited configuration-field ordering while keeping the fix localized to the affected caller. No shared configuration design or library architecture is changed.
• API: No public API, export, default, or configuration contract is modified. The existing scene.num_envs value is used to produce per-environment joint targets, consistent with the surrounding per-environment state handling.
• Implementation: Both action templates are created on the simulation device and repeated to shape [scene.num_envs, 2] once before the simulation loop, preserving reuse across steps without introducing per-step allocation. The single-environment behavior remains equivalent while multi-environment runs receive one target row per environment.

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.

[greptile-apps]

Greptile Summary

This PR repairs the add_new_robot tutorial for startup and multi-environment execution.

• Passes num_envs by keyword to avoid dependence on inherited dataclass field order.
• Expands both Jetbot wheel-action templates to the scene’s environment count.

Confidence Score: 5/5

The PR appears safe to merge, with both tutorial failures corrected consistently with the underlying configuration and actuator APIs.

Keyword construction correctly populates the inherited scene fields, while repeating the two-wheel command rows produces the exact batch shape required for every cloned Jetbot.

Important Files Changed

Filename	Overview
scripts/tutorials/01_assets/add_new_robot.py	The scene configuration now binds num_envs correctly, and the wheel targets match the setter’s required (num_envs, 2) shape.

_{Reviews (1): Last reviewed commit: "Drop the standalone tutorial CI jobs" | Re-trigger Greptile}

[isaaclab-bot]

Backported to release/3.0.0 as adf8d9c.