Encodes an observation into its latent representation.

Select action for inference/evaluation

Get all recorded debug steps. Returns: List of all DebugStep objects (may be empty if disabled).

partition `lst` into that the mininal number of groups that as evenly sized as possible and are at most `max_group_size` in size

Moves the processor's normalization stats to the specified device. Args: device: The target PyTorch device. Ret

Write calibration parameters to the motors.

Find stage and within-stage progress (tau) for a frame. Args: current_frame: Frame index relative to episode start episode_length

Return the most recent frame captured immediately (Peeking). This method is non-blocking and returns whatever is currently in the mem

Save model, optim, and other training state to a local or remote directory.

Read a value from multiple motors.

Builds a "Write" instruction. The parameters for Write are: param[0] = start_address L param[1] = start_

returns the latest is checkpoint directory in `checkpoint`, returns `checkpoint` if it is already a checkpoint dir

Disable torque on selected motors.

Compute the loss for the given model Args: batch: Dictionary containing: - action: Action tensor

Wait for `seconds` with better precision than time.sleep alone at the expense of more CPU usage. Parameters: - seconds: duration to wa

Capture and return a single frame from the camera synchronously. This is a blocking call that will wait for the hardware and its SDK.

Write a value to a single motor. Positions are always in degrees. Can write 'Goal_Position', 'Kp', or 'Kd'.

Aggregates and filters pipeline features to create a dataset-ready features dictionary. This function transforms initial features using the

Compute comprehensive statistics for array features along specified axes. This function calculates min, max, mean, std, and quantiles (1%, 10%, 5

Convert hardware-specific features to a LeRobot dataset feature dictionary. This function takes a dictionary describing hardware outputs (like jo

List the contents of a local or remote directory. If ``recurse=False``, only the immediate children of the directory are returned, otherwise

Fixture to provide the PI0Fast preprocessor for tests.

Close environment and disconnect robot.

Start the timer.

Heuristic to estimate the number of samples based on dataset size. The power controls the sample growth relative to dataset size. Lower the po

Get the local rank per filesystem, meaning that, regardless of the number of nodes, if all ranks share the same filesystem then `get_fs_local_rank

Reset environment to initial state. Args: seed: Random seed for reproducibility. options: Additional reset options.

Update chunk and file size settings after dataset creation. This allows users to customize storage organization without modifying the constru

Get motor name from name or ID.

Add new data to the buffer, which could potentially mean shifting old data out. The new data should contain all the frames (in order) of any

Add multiple features to a LeRobotDataset in a single pass. This is more efficient than calling add_feature() multiple times, as it only copi

Write implementation-specific recommended settings to every motor. Typical changes include shortening the return delay, increasing ac

Feed a frame to the encoder for a specific camera. A copy of the image is made before enqueueing to prevent race conditions with came

Create a RobotClientConfig from a dictionary.

Get the world size of the default distributed process group. .. warning:: This will always return 1 if a distributed group has not b

Instantiates a policy configuration object based on the policy type. This factory function simplifies the creation of policy configuration o

Update controller state - call this once per frame.

Find the first object logged under a given key.

Delete episodes from a LeRobotDataset and create a new dataset. Args: dataset: The source LeRobotDataset. episode_indices: List o

Format video tracks start_end_only: Only track the first and last point of each object first_object: Only track the first ob

Get current chunk and file size settings. Returns: Dict containing chunks_size, data_files_size_in_mb, and video_files_size_in_mb

Initialize weights of a linear or embedding module. :param config: The model config. :param module: The linear or embedding submodule to

Load model, optim, and other training state from a local or remote checkpoint directory created via :meth:`save()` or :meth:`save_asy

Creates the pre- and post-processing pipelines for the ACT policy. The pre-processing pipeline handles normalization, batching, and device placem