PalmDex: A Multimodal Robotic Manipulation Dataset with Tactile Sensing

Rimbot's PalmDex is a multimodal robotic manipulation dataset collected via human teleoperation. It features synchronized dual-camera video, dual-hand tactile sensing, and hand pose tracking across diverse real-world environments, with action-segment level annotations. The dataset is described as a living collection, last updated on 2026-06-23.

Use Cases

• Train multimodal robotic control models based on synchronized video and tactile data.
• Develop action segmentation algorithms based on the annotated action-segment level labels.
• Research human-like dexterous manipulation based on teleoperated demonstrations.
• Benchmark tactile sensing and pose tracking models based on the dual-hand sensor data.

Strengths

• Multimodal data includes synchronized dual-camera video, dual-hand tactile sensing, and hand pose tracking.
• Demonstrations are collected via human teleoperation without a specific robot embodiment, which may increase generality.
• Data is annotated at the action-segment level with categorical labels.

Limitations

• Row count, file formats, and column-level documentation are unknown, which may limit suitability assessment.
• The dataset is described as 'living'; its current scope and completeness are not specified.
• License information is unknown, which may restrict usage.

Provenance

Source

Rimbot on Hugging Face.

Collection Method

Collected via human teleoperation.

Time Range

null

Freshness

Last updated 2026-06-23 02:02:49.

Geography

null