TrajectoryResampling

class standard_e2e.dataset_utils.augmentation.TrajectoryResampling(history_target_timestamps=None, future_target_timestamps=None, target_frame_names=None)

Bases: FrameAugmentation

Resample past/future/preference trajectories onto target timestamps.

Parameters:

• history_target_timestamps (list[float] | None)

• future_target_timestamps (list[float] | None)

• target_frame_names (list[str] | None)

ALLOWED_REGIMES = ['train', 'val', 'test']

augment(frames, regime)

Apply augmentation to the given frame data.

Parameters:

• frames (dict[str, TransformedFrameData]) – The frame data to augment.

• regime (str) – The regime for which the augmentation is applied.

Returns:

The augmented frame data.

Return type:

dict[str, TransformedFrameData]