reference/generated/torchrl.render.play_mujoco_wasm_trajectory

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

play_mujoco_wasm_trajectory

torchrl.render.play_mujoco_wasm_trajectory(qpos: Any, *, dt: float | None = None, fps: float | None = None, port: int = 5178, viewer_origin: str | None = None, degrees: bool = False, pause: bool = True, loop: bool = False, interpolate: bool = True, wait: bool = False, timeout: float | None = None, viewer_dir: str | Path | None = None) → dict[str, Any] | None

Plays a qpos trajectory in a live MuJoCo WASM notebook viewer.

Parameters:

• qpos – Iterable of waypoints, each containing qpos values.

• dt – Seconds between waypoints. Defaults to 0.1 when fps is not supplied.

• fps – Optional playback rate in frames per second.

• port – Viewer port when viewer_origin is not supplied.

• viewer_origin – Browser origin of the viewer iframe.

• degrees – Whether the provided values are degrees.

• pause – Whether physics should remain paused after playback.

• loop – Whether the browser should loop the trajectory.

• interpolate – Whether to linearly interpolate between waypoints.

• wait – Whether to block until a non-looping trajectory completes.

• timeout – Seconds to wait before raising TimeoutError.

• viewer_dir – Optional generated viewer directory. When provided, the trajectory is written under the viewer’s public/ directory and a viewer iframe is displayed with an autoplay URL. This avoids relying on notebook JavaScript outputs in JupyterLab.

Returns:

The acknowledgement payload when wait=True. When viewer_dir is provided, returns a small metadata dictionary for the generated autoplay trajectory.

Examples

>>> from torchrl.render.mujoco_wasm import play_mujoco_wasm_trajectory
>>> callable(play_mujoco_wasm_trajectory)
True