Tactile-Based Robotic Peg Extraction Dataset

This dataset provides robotic and tactile sensor data captured using two multi-modal tactile sensing (BioIn-Tacto [1, 2]) modules mounted on the end-effector of an OpenManipulator X. The sensor includes barometric and MARG (Magnetic, Angular Rate, and Gravity) data to support research in tactile manipulation. The dataset was collected in teleoperation experiments involving the extraction of differently shaped pegs from a base with holes using a robotic manipulator arm. The total number of extraction episodes in the dataset is 96. The dataset also includes Reinforcement Learning pre-trained data. The dataset can be used to pre-train a reinforcement learning model to perform peg-in-hole tasks and to study how pre-training affects a manipulator’s ability to infer tactile signals and improve success rates of the manipulator.

The data is organized into folders representing each object runs:

Data/
└RL/
│ └Object<1|2|3|>
│ └recordstep_<timestr>_object<1|2|3>_pretained_<ID>.csv
└Teleop/
├csv2dataframe.py
├README.md
└──csv/
└Object<1|2|3|>/
└robot_O<1|2|3>_T<n>_A<0|45|90|135|180>_<timestr>/
├imu1_data_raw.csv
├imu1_mag.csv
├imu2_data_raw.csv
├imu2_mag.csv
├imu_data1.csv
├imu_data2.csv
├joint_states.csv
├m_baros_serial1.csv
├m_baros_serial2.csv
├pressure_viz_left.csv
├pressure_viz_right.csv
├raw_barometers_teensy1.csv
├raw_barometers_teensy2.csv
├raw_imu_teensy1.csv
├raw_imu_teensy2.csv
├robot_instruction.csv
├tf.csv
└tf_static.csv

- csv2dataframe.py: Converts the data into dataframes
- Object<1|2|3|>/: Three folders with data from each object
- recordstep_<timestr>_object<1|2|3>_pretained_<ID>: Files with RL data for each object.
- robot_O<1|2|3>_T<n>_A<0|45|90|135|180>_<timestr>/: Data dollected from each object at different angles.

[1] T. E. Alves de Oliveira, A. -M. Cretu and E. M. Petriu, "Multimodal Bio-Inspired Tactile Sensing Module," in IEEE Sensors Journal, vol. 17, no. 11, pp. 3231-3243, 1 June1, 2017, https://doi.org/10.1109/JSEN.2017.2690898.
[2] T. E. Alves de Oliveira, V. Prado da Fonseca, BioIn-Tacto: A compliant multi-modal tactile sensing module for robotic tasks, HardwareX, Volume 16, 2023, e00478, ISSN 2468-0672, https://doi.org/10.1016/j.ohx.2023.e00478.

本数据集收录了采用开放式机械臂X的末端执行器上安装的双模态触觉感知（BioIn-Tacto [1, 2]）模块所采集的机器人与触觉传感器数据。该传感器集成了气压和MARG（磁场、角速度和重力）数据，旨在支持触觉操作研究。数据集通过遥操作实验收集，实验内容为使用机器人操作臂从带孔底座中提取不同形状的销钉。数据集中提取场景的总次数为96次。此外，数据集还包含了强化学习预训练数据。该数据集可用于预训练强化学习模型以执行销钉插入孔洞任务，并研究预训练如何影响操作者对触觉信号的推断能力及其成功率的提升。

数据组织结构如下：

Data/
├── RL/
│ └── Object<1|2|3|>/
│ └── recordstep_<timestr>_object<1|2|3>_pretained_<ID>.csv
└── Teleop/
├── csv2dataframe.py
├── README.md
└── csv/
└── Object<1|2|3|>/
└── robot_O<1|2|3>_T<n>_A<0|45|90|135|180>_<timestr>/
├── imu1_data_raw.csv
├── imu1_mag.csv
├── imu2_data_raw.csv
├── imu2_mag.csv
├── imu_data1.csv
├── imu_data2.csv
├── joint_states.csv
├── m_baros_serial1.csv
├── m_baros_serial2.csv
├── pressure_viz_left.csv
├── pressure_viz_right.csv
├── raw_barometers_teensy1.csv
├── raw_barometers_teensy2.csv
├── raw_imu_teensy1.csv
├── raw_imu_teensy2.csv
├── robot_instruction.csv
└── tf.csv

csv2dataframe.py: 将数据转换为数据框
Object<1|2|3|>: 包含每个物体数据的三个文件夹
recordstep_<timestr>_object<1|2|3>_pretained_<ID>: 每个物体的强化学习数据文件
robot_O<1|2|3>_T<n>_A<0|45|90|135|180>_<timestr>/: 在不同角度下收集的每个物体的数据。

[1] T. E. Alves de Oliveira, A. -M. Cretu 和 E. M. Petriu, “多模态生物启发式触觉感知模块,” 发表于《IEEE传感器杂志》, 第17卷，第11期，第3231-3243页，2017年6月1日，https://doi.org/10.1109/JSEN.2017.2690898.
[2] T. E. Alves de Oliveira, V. Prado da Fonseca, BioIn-Tacto: 一种适用于机器人任务的合规多模态触觉感知模块, HardwareX, 第16卷，2023年，e00478，ISSN 2468-0672，https://doi.org/10.1016/j.ohx.2023.e00478.