Tactile, Audio, and Visual Dataset During Bare Finger Interaction with Textured Surfaces

Area of Science:

• Neuroscience
• Robotics
• Human-Computer Interaction

Background:

• Human tactile perception relies on integrating multiple sensory inputs.
• Existing datasets often use artificial probes, limiting naturalistic data acquisition.
• A comprehensive dataset of natural human tactile exploration is needed.

Purpose of the Study:

• To create a multi-modal dataset of human tactile exploration.
• To provide a resource for studying multi-sensory integration in touch.
• To support the development of artificial tactile perception systems.

Main Methods:

• Recorded concurrent haptic, audio, and visual signals from 10 participants interacting with 10 textured surfaces using bare fingers.
• Collected stereoscopic images, fingertip kinematics, applied load, sound, and vibrations.
• Included subjective evaluations from 15 additional participants via questionnaires.
• Employed both controlled and free exploration protocols.

Main Results:

• The dataset captures naturalistic haptic data, unlike studies using rigid probes.
• Preliminary analysis shows promising accuracy in surface identification using multi-modal classifiers.
• Demonstrates the potential for advancing texture recognition algorithms.

Conclusions:

• The dataset is a valuable resource for research in human tactile perception and multi-sensory integration.
• Facilitates the development of advanced algorithms for artificial tactile perception.
• Supports the creation of more sophisticated human-machine interfaces.