Dataset Viewer
The dataset viewer is not available for this dataset.
Unexpected token '<', "<html> <h"... is not valid JSON

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

AIDAS_franka_pnp_100 (TsFile)

TsFile conversion of the LeRobot-style Franka pick-and-place dataset broxKIM/AIDAS_franka_pnp_100.

  • Original dataset: https://huggingface.co/datasets/broxKIM/AIDAS_franka_pnp_100
  • Author: broxKIM
  • Robot type: franka
  • Created with: LeRobot-style metadata (codebase_version v3.0)
  • Modalities: Time-series
  • Dropped / not included: the source timestamp column (redundant with Time) and the camera video streams (not uploaded; see Conversion Notes).

Dataset Description

This dataset contains Franka pick-and-place demonstrations in a single train split.

Robot type franka
Tasks 4 (task_index=0..3)
Episodes 98
Frames / TsFile rows 60,918
Sampling rate 15 fps
Original cameras observation.images.cam_front, observation.images.cam_wrist

Each frame records Cartesian state, gripper state, joint state, target state, the combined 20-dimensional observation state, and the 7-dimensional action.

Conversion Notes

Converted to TsFile with the generic LeRobot converter.

  • One TsFile for the whole train split. All 98 episodes are stored in data/aidas_franka_pnp_100.tsfile. The source columns episode_index and task_index are TAG columns (the TsFile device dimension). Query one episode with WHERE episode_index=N, or one task with WHERE task_index=K.
  • Vector columns flattened to single-precision FLOAT, preserving the source column name (. -> _, element index appended):
    • observation.state.cartesian [6] -> observation_state_cartesian_0 .. observation_state_cartesian_5
    • observation.state.joints [7] -> observation_state_joints_0 .. observation_state_joints_6
    • observation.state.target [6] -> observation_state_target_0 .. observation_state_target_5
    • observation.state [20] -> observation_state_0 .. observation_state_19
    • action [7] -> action_0 .. action_6
  • Scalar state kept. observation.state.gripper is renamed to observation_state_gripper; index is renamed to sample_index.
  • Time. Time = round(timestamp * 1000) ms, restarting per episode. The source timestamp column is dropped because it equals Time / 1000 seconds. frame_index is kept.
  • Videos are NOT included in this repository. The original camera streams remain in the source dataset: https://huggingface.co/datasets/broxKIM/AIDAS_franka_pnp_100/tree/main/videos
  • meta/ is mirrored from the source with meta/info.json rewritten to describe the converted TsFile schema and video policy. Aside from the redundant timestamp column noted above, no numeric time-series rows are dropped.

Schema

Column Role Type
Time TIME (ms) INT64
episode_index TAG INT64
task_index TAG INT64
frame_index FIELD INT64
sample_index FIELD INT64
observation_state_gripper FIELD FLOAT
observation_state_cartesian_0 .. observation_state_cartesian_5 FIELD FLOAT
observation_state_joints_0 .. observation_state_joints_6 FIELD FLOAT
observation_state_target_0 .. observation_state_target_5 FIELD FLOAT
observation_state_0 .. observation_state_19 FIELD FLOAT
action_0 .. action_6 FIELD FLOAT

Read Example

# Read data/aidas_franka_pnp_100.tsfile with the Apache TsFile SDK.
Downloads last month
137