README.md

metadata

license: cc-by-4.0
pretty_name: TexasPokerRobot Small
tags:
  - robotics
  - imitation-learning
  - visuomotor-control
  - npz
size_categories:
  - n<1K

TexasPokerRobot Small

This repository is a small public subset of Winniechen2002/TexasPokerRobot. It is intended as the compact dataset release for NeurIPS review requirements.

The subset contains one original .npz trajectory file, data_0001.npz, from each of the 14 task folders in the full dataset. The files are not modified or repacked; the directory names and archive contents follow the source dataset.

Contents

• 14 .npz files total
• 1 file per task
• Approximately 4.38 GiB total
• Source revision: main of Winniechen2002/TexasPokerRobot

Tasks included:

• pick_up_left
• pick_up_right
• pull_5
• pull_10
• pull_50
• pull_100
• push_5
• push_10
• push_50
• push_100
• put_down_left
• put_down_right
• show_left
• show_right

Each archive uses the same fields as the full dataset. A representative file contains:

• joint_positions
• joint_velocities
• joint_efforts
• images_cam0
• images_cam1
• images_cam2
• depth_cam0
• depth_cam1
• depth_cam2

See manifest.csv for the exact file list and byte sizes.

Loading Example

import numpy as np

episode = np.load("pick_up_left/data_0001.npz", allow_pickle=True)
print(episode.files)

Some arrays may require allow_pickle=True, matching the original .npz serialization.

License

This subset follows the source dataset license: CC BY 4.0.