Pick Place Tactile: Robot Manipulation Data with Tactile Sensing

A dataset of 217 episodes and 115,245 frames for robot manipulation tasks, created using the LeRobot framework. The data was collected using a Unitree G1 Inspire robot equipped with two cameras and includes tactile sensing information. The dataset was uploaded by dishantpatel1207 and last updated in April 2026.

Use Cases

• Training imitation learning models based on recorded robot episodes and frames.
• Developing reinforcement learning policies for pick-and-place tasks.
• Researching multimodal perception combining visual and tactile data.
• Benchmarking robot manipulation algorithms on a defined task.

Strengths

• Contains 217 episodes, providing a substantial number of task demonstrations.
• Includes 115,245 frames at 30 FPS, offering high temporal resolution.
• Data was collected with a specific robot model (Unitree_G1_Inspire_2cam), ensuring hardware consistency.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset description is limited; actual data quality requires manual inspection after download.

Provenance

Source

huggingface

Collection Method

Created using the LeRobot framework, likely involving teleoperation or autonomous recording.

Time Range

null

Freshness

Last updated 2026-04-10 12:52:03; freshness should be verified.

Geography

null