real01b_square_d2_r4_quad_baseline (TsFile)

Apache TsFile version of ankile/real01b-square-d2-r4-quad-baseline-nocf-freecf-iql-s2final-n16-heval-sobolseed2026070304.

Overview

This dataset is a LeRobot v3.0 evaluation dataset of Franka policy rollouts on a real01b square-routing task (quad-baseline, D2 R4).

• Robot: Franka (franka)
• Scale: 200 episodes, 56,894 frames, 15 fps
• Split: train
• Cameras (not uploaded): wrist_left, wrist_right, side_1, side_2

Schema (TsFile structure)

The TsFile table is named real01b_square_d2_r4_quad_baseline.

Role	Columns
Time	Time, INT64 milliseconds
TAG	episode_index, task_index
FIELD	frame_index, sample_index
FIELD	observation_state_0 … observation_state_6, FLOAT
FIELD	action_0 … action_6, FLOAT
FIELD	action_cartesian_velocity_0 … action_cartesian_velocity_5, FLOAT
FIELD	action_cartesian_position_0 … action_cartesian_position_5, FLOAT
FIELD	action_joint_velocity_0 … action_joint_velocity_6, FLOAT
FIELD	action_joint_position_0 … action_joint_position_6, FLOAT
FIELD	observation_state_cartesian_position_0 … observation_state_cartesian_position_5, FLOAT
FIELD	observation_state_joint_position_0 … observation_state_joint_position_6, FLOAT
FIELD	observation_state_joint_velocity_0 … observation_state_joint_velocity_6, FLOAT
FIELD	observation_state_cartesian_velocity_0 … observation_state_cartesian_velocity_5, FLOAT
FIELD	telemetry_franka_motor_torques_external_0 … telemetry_franka_motor_torques_external_6, FLOAT
FIELD	telemetry_franka_ee_wrench_0 … telemetry_franka_ee_wrench_5, FLOAT
FIELD	telemetry_franka_motor_torques_measured_0 … telemetry_franka_motor_torques_measured_6, FLOAT
FIELD	telemetry_franka_joint_torques_computed_0 … telemetry_franka_joint_torques_computed_6, FLOAT
FIELD	steps_to_go
FIELD	source
FIELD	intervention
FIELD	success
FIELD	is_valid
FIELD	reward
FIELD	done
FIELD	policy_id
FIELD	round_id
FIELD	action_gripper_position
FIELD	action_gripper_velocity
FIELD	observation_state_gripper_position
FIELD	manifest_idx
FIELD	nut_x
FIELD	nut_y
FIELD	nut_yaw
FIELD	peg_x
FIELD	peg_y

Vector columns are flattened into scalar FLOAT measurements (single precision). The source index column is retained as sample_index.

Conversion notes

• Time = round(timestamp * 1000) with millisecond precision; time restarts inside each episode, while episode_index and task_index identify the TsFile device.
• The original timestamp field is omitted because it is exactly represented by Time / 1000.
• Camera video streams are NOT included in this repository; they remain in the source videos tree.
• meta/ is mirrored from the source. Aside from the redundant timestamp column and the excluded videos, no source rows or numeric fields are dropped.

Read example

from tsfile import TsFileReader

with TsFileReader("data/real01b_square_d2_r4_quad_baseline.tsfile") as reader:
    print(reader.get_all_table_schemas()["real01b_square_d2_r4_quad_baseline"])

Source & license

• Original dataset: https://huggingface.co/datasets/ankile/real01b-square-d2-r4-quad-baseline-nocf-freecf-iql-s2final-n16-heval-sobolseed2026070304
• Author: ankile
• License: apache-2.0