Forward to each term's :meth:`RecorderTerm.record_post_reset`.

Recreate the render context after model fields are expanded. Called by Simulation.expand_model_fields() for domain randomization.

Reflect rotary motor properties through a transmission to linear joint properties. Converts motor specs from rotary coordinates (rad, rad/s, Nm) to

Generate and render a single terrain type, return RGB array.

Reset bias values for specified environments.

Zero actuator targets and reset actuator internal state. Called by the scene when environments are reset at episode boundaries, and by comman

Extend BaseViewer.reset_environment to clear reward and metrics histories.

Reset root state for floating-base or mocap fixed-base entities. For floating-base entities: Resets pose and velocity via write_root_state_to_sim()

Reset to exact reference state at a specific frame. Like ``_resample_command`` but deterministic: no random perturbations to pose, velocity,

Run the analysis comparing instantaneous vs history contact detection.

Read instantaneous contact at the end of each policy step.

Launch a Viser viewer showing the bouncing ball with contact forces.

Read full substep history to catch mid-decimation contacts.

Sample using gaussian distribution. Args: mean: Mean of the gaussian. std: Std of the gaussian. size: The shape of the te

r"""Sample using log-uniform distribution within a range. The log-uniform distribution is defined as a uniform distribution in the log-space. It

Save benchmark results to JSON, appending to existing data.

Set site effort targets. Args: effort: Target site efforts with shape (N, num_sites). site_ids: Optional site indices to set. If None

Set tendon effort targets. Args: effort: Target tendon efforts with shape (N, num_tendons). tendon_ids: Optional tendon indices to se

Set tendon length targets. Args: length: Target tendon lengths with shape (N, num_tendons). tendon_ids: Optional tendon indices to se

Set tendon velocity targets. Args: velocity: Target tendon velocities with shape (N, num_tendons). tendon_ids: Optional tendon indice

Setup the viewer GUI and scene.

Create camera feed controls under the active GUI folder.

Create rewards/metrics tabs based on available managers.

Resolves the module and function names to return the function. Args: name: The function name. The format should be 'module:attribute_name

Test that mjlab can create and close an environment.

Unpack packed uint32 RGB data into separate channels. Called from Simulation._sense_kernel() so it gets captured in the CUDA graph. No-op if

Update term plots from the selected environment.

Update sensor state after a simulation step. Invalidates the data cache. Override in subclasses that need per-step updates, but call `super()

Update the environment origins based on the terrain levels.

Update scene and sync debug visualizations.

Context manager that captures state before and checks for NaN/Inf after. Usage: with nan_guard.watch(wp_data): mjwarp.step(wp_model

r"""Wraps input angles (in radians) to the range :math:`[-\pi, \pi]`. This function wraps angles in radians to the range :math:`[-\pi, \pi]`, suc

Set the root COM velocity into the simulation. Args: root_velocity: Tensor of shape (N, 6) where N is the number of environments. C

Write the MjSpec to disk. Operates on a copy of the spec to avoid mutating the original.

Apply noise and additive bias to input data.

Returns the value of the term required by the manager.

Tick impulse state: expire old impulses, trigger new ones. Args: env: The environment instance. env_ids: Unused (step events always o