qpos of the articulated (non-free) joint(s) for ``region``. A multi-drawer cabinet has one joint per drawer (``<obj>_<level>_level``); a regi

Tier-1 task files (Tier-3 exempt) that define a class method named ``method``.

Iterations on which the delayed update fires, for the buggy vs fixed gate.

For compatibility with Isaac Lab native RSL-RL env wrapper.

True only when the installed ``libero`` is actually LIBERO-plus (>130 tasks).

Import a ``get_started/NN_*.py`` example module by file path. The module is registered in ``sys.modules`` before execution so that the ``@dat

Resolve the MotionCommand(Manager) or return None.

``(N,3)`` world position of a scene *object* on the Newton path.

Two-stage planetary reflected inertia (mjlab.utils.actuator).

Return per-body (pos, quat, lin_vel, ang_vel) for all tracked robot bodies. MuJoCo path; falls back to identity for Newton (caller will see

Composite ``overlay_img`` behind the foreground, matching SimplerEnv's get_obs. Args: color: rendered RGB, shape (H,W,3) or (H,W,4); only

Body-frame angular velocity of the asset's base. Shape ``(num_envs, 3)``.

Create a SharedNDArray directly from a SharedMemoryManager Parameters ---------- mem_mgr SharedMemoryManager insta

Generate point cloud using depth image only. Input: depth: [numpy.ndarray, (H,W), numpy.float32] depth image

Ensure environment is in clean initial state with intelligent validation.

Convert a single (w, x, y, z) quaternion → Euler angles (roll, pitch, yaw) in radians. Input shape: (4,), returns a tuple of scalars.

Extract state dictionary from handler states. Args: env: Environment with handler scenario: Scenario configuration to get joint n

Extract state dictionary from TensorState observation. Args: obs: TensorState observation scenario: Scenario configuration to get

Extract states from observation tensor.

Extract states from observation tensor.

Extract states from observation tensor.

Extract states from observation tensor. Args: obs: TensorState observation handler: Simulator handler key: "objects" or "

Extract states from observation tensor. Args: obs: TensorState observation handler: Simulator handler key: "objects" or "

Find bodies in the articulation based on the name keys. Please check the :meth:`isaaclab.utils.string_utils.resolve_matching_names` function for

Look up a benchmark task spec by canonical name or alias.

Get total number of completed episodes.

Get geometric center of the object.

Get average return and length over recent episodes.

Calculate value from logits using support

Calculate value from logits using support

Keep points in workspace as input. Input: cloud: [np.ndarray, (H,W,3), np.float32] scene point cloud

Move the robot to the target pose.

Convert a legacy .msh file to the .obj format.

Add ``<position>`` actuators to each named joint, return path to patched XML. Also sets MuJoCo solver options to mjlab defaults — most critically

Pattern match the sub_names to all_names using regex.

Pattern match the sub_names to all_names using regex.

Convert quaternion [w, x, y, z] to RPY angles [roll, pitch, yaw].

Convert quaternion [w, x, y, z] to RPY angles [roll, pitch, yaw].

Convert quaternion [w, x, y, z] to RPY angles [roll, pitch, yaw].

Convert pytorch3d quaternion (w,x,y,z) to scipy format (x,y,z,w).

Reshape selected dimensions in a tensor to a target dimension. Args: x (torch.Tensor): tensor to reshape begin_axis (int): b

Return the robot-specific teleoperation profile.

Seed will be randomly initialized if it's None.

Transform points to new coordinates with transformation matrix. Input: cloud: [np.ndarray, (N,3), np.float32] po

Map RoboTwin's 14-D bimanual vector onto the embodiment's joint targets. ``vec`` is ``[L_arm(6), L_grip, R_arm(6), R_grip]``. Gripper values are