Cache-SCA/SO101-DualArm-teleop_fold_towel_100epi_10fps
SO101 DualArm Teleop Fold Towel 100epi 10fps LeRobot v3.0 derivative of CoRL2026-CSI/SO101-DualArm-teleop_fold_towel_100epi. Robot: bi_so_follower Task instruction: Fold the towel in half. Episodes: 100 Frames: 21,749 FPS: 10 Cameras: observation.images.left_wrist, observation.images.left_top, observation.images.right_wrist Action keys: action Derivation This dataset downsamples the source dataset from 30 fps to 10 fps by keeping every 3rd source frame. Source… See the full description on the dataset page: https://huggingface.co/datasets/Cache-SCA/SO101-DualArm-teleop_fold_towel_100epi_10fps.
SO101 DualArm Teleop Fold Towel 100epi 10fps
LeRobot v3.0 derivative of CoRL2026-CSI/SO101-DualArm-teleop_fold_towel_100epi.
- Robot:
bi_so_follower - Task instruction: Fold the towel in half.
- Episodes: 100
- Frames: 21,749
- FPS: 10
- Cameras:
observation.images.left_wrist,observation.images.left_top,observation.images.right_wrist - Action keys:
action
Derivation
This dataset downsamples the source dataset from 30 fps to 10 fps by keeping every 3rd source frame.
Source frame mapping within each episode:
output_frame_index k -> source_frame_index k * 3
output_timestamp -> k / 10For each retained source frame i, all observation, image, state, action, and task columns are copied from source frame i. Action labels are not relabelled from future states and are not shifted. This preserves the source dataset's causal LeRobot row pairing for the retained frames.
Contents
Required state/action columns are preserved:
observation.stateaction
All RGB video streams are preserved and re-encoded at 10 fps:
observation.images.left_topobservation.images.left_wristobservation.images.right_wrist
Provenance
- Source dataset:
CoRL2026-CSI/SO101-DualArm-teleop_fold_towel_100epi - Derived dataset:
CoRL2026-CSI/SO101-DualArm-teleop_fold_towel_100epi_10fps - Source FPS: 30
- Target FPS: 10
- Frame stride: 3
- Source episodes: 0-99
- Generated with:
scripts/tools/dataset/derive_subset_fps.py
