Instantiate a policy model. This factory function handles the logic of creating a policy, which requires determining the input and outpu

Constructs pre-processor and post-processor pipelines for the SAC policy. The pre-processing pipeline prepares input data for the model by:

Constructs pre-processor and post-processor pipelines for the TDMPC policy. The pre-processing pipeline prepares input data for the model by

Constructs pre-processor and post-processor pipelines for the VQ-BeT policy. The pre-processing pipeline prepares input data for the model b

Return the p-th percentile of recorded times.

Split a LeRobotDataset into multiple smaller datasets. Args: dataset: The source LeRobotDataset to split. splits: Either a dict m

Compute τ_t = (t - s_k) / (e_k - s_k) ∈ [0, 1]. Returns 1.0 for zero-duration subtasks.

Create a default ACT configuration for testing.

Create a default VQBeT configuration for testing.

Create a default SAC configuration for testing.

Create a default TDMPC configuration for testing.

Extracts the subtask from the transition's complementary data. Args: transition: The environment transition. Re

Initializes the Rerun SDK for visualizing the control loop. Args: session_name: Name of the Rerun session. ip: Optional IP f

Constructs pre-processor and post-processor pipelines for the reward classifier. The pre-processing pipeline prepares input data for the cla

Constructs pre-processor and post-processor pipelines for a diffusion policy. The pre-processing pipeline prepares the input data for the mo

Modify tasks in a LeRobotDataset. This function allows you to either: 1. Set a single task for the entire dataset (using `new_task`) 2. S

Plot trajectories across multiple dimensions. This function plots a tensor's values across time for multiple dimensions, with each di

Assign the correct ID and baud-rate to a single motor. This helper temporarily switches to the motor's current settings, disables torque, set

Build codec-specific options dict for video encoding.

Return the expected ID found in feedback payload byte0 for this motor. Robstride MIT feedback frames encode an ID in data[0]. Some setups exp

Convert motor specification to list of motor names.

Create a TimedObservation with a given state vector.

Merge LeRobot grouped feature dicts. - For 1D numeric specs (dtype not image/video/string) with "names": we merge the names and recompute the sha

Create a default Diffusion configuration for testing.

Create a default Reward Classifier configuration for testing.

Create default dataset statistics for testing.

Create default dataset statistics for testing.

Create default dataset statistics for testing.

Create default dataset statistics for testing.

Creates the initial features dict for the dataset from action and observation specs. Args: action: A dictionary of action feature na

RTC guidance wrapper around an existing denoiser. This method wraps an original denoising callable that only takes ``x_t`` and return

Finish encoding the current episode. Sends sentinel values, waits for encoder threads to complete, and collects results. Ret

Get all debug steps from tracker. Returns empty list if tracker is disabled or None.

Replaces normalization statistics in an existing `PolicyProcessorPipeline` instance. This function creates a deep copy of the provided pipel

Check if the package spec exists and grab its version to avoid importing a local directory. Args: pkg_name: The name of the package

Helper to create SubtaskAnnotation from list of (name, start_sec, end_sec).

Makes a gym vector environment according to the config or Hub reference. Args: cfg (EnvConfig | str): Either an `EnvConfig` object descri

Registers a function to be called before each step. Args: fn: A callable that accepts the step index and the current transition.

Validate vcodec and resolve 'auto' to best available HW encoder, fallback to libsvtav1.

Start encoder threads for a new episode. Args: video_keys: List of video feature keys (e.g. ["observation.images.laptop"])

Execute one environment step with given action.

Check if the cached dataset contains all requested episodes and their video files.

Get CAN ID for a motor.

Get motor recv_id from name or ID.

Check if directory has JSON files but no processor configs. This method implements smart migration detection to avoid false positives:

Return the most recent new frame. This method retrieves the latest frame captured by the background thread. If a new frame is already

Create default dataset statistics for testing.

Create default dataset statistics for testing.

Loads the JSON data from `fpath` and recursively fills `obj` with the corresponding values (strictly matching structure and types). Tuple

Flatten a nested dictionary by joining keys with a separator. Example: >>> dct = {"a": {"b": 1, "c": {"d": 2}}, "e": 3} >>> print

Format points for a single image