A Bionic Textile Sensory System for Humanoid Robots Capable of Intelligent Texture Recognition

Area of Science:

• Robotics
• Materials Science
• Artificial Intelligence

Background:

• Advanced prosthetics and humanoid robots require artificial tactile perception systems.
• Current systems often integrate separate sensors for slow adaptive (SA) and fast adaptive (FA) mechanoreceptors.
• Simultaneously perceiving static and dynamic forces for texture recognition in a single device is a significant challenge.

Purpose of the Study:

• To develop a high-performance artificial tactile perception system.
• To create a single-device sensor capable of perceiving both static and dynamic forces.
• To enable accurate surface-texture recognition for robotics and prosthetics.

Main Methods:

• A textile pressure/tactile (PT) sensor was developed based on the piezoresistive principle.
• The sensor features a 3D multiscale architecture with a honeycomb-like sensing fabric.
• An automated surface-texture-recognition system was built integrating the PT sensor, machine learning, and robotic components.

Main Results:

• The PT sensor demonstrated high sensitivity and rapid response to both vibrations and static forces.
• The integrated system achieved high recognition accuracy (>98.9%) for classifying unknown textiles.
• The system successfully differentiated fine surface textures across various scanning speeds (50-300 mm s⁻¹).

Conclusions:

• The developed textile PT sensor offers a promising solution for artificial tactile perception.
• This technology advances the development of sophisticated prosthetics and humanoid robots.
• The findings support the future creation of interactive artificial intelligence systems with enhanced sensory capabilities.