Artificial Tactile Sensory Finger for Contact Pattern Identification Based on High Spatiotemporal Piezoresistive...

Area of Science:

• Robotics
• Materials Science
• Biomimetics

Background:

• Human tactile perception relies on specialized receptors for pattern recognition.
• Existing artificial tactile sensors struggle to match human fingertip spatiotemporal resolution and accuracy.
• Integrating sensors with machine learning is key but challenging for high-fidelity tactile sensing.

Purpose of the Study:

• To develop an artificial tactile finger capable of high spatiotemporal resolution contact pattern identification.
• To emulate and potentially surpass human fingertip tactile sensing capabilities.
• To create a versatile platform for robotic tactile sensing applications.

Main Methods:

• Developed an artificial tactile finger integrating a high spatiotemporal piezoresistive sensor array (PRSA).
• Utilized a convolutional neural network (CNN) model for pattern recognition.
• Compared artificial finger performance against human tactile perception using embossed and curved patterns.

Main Results:

• The artificial finger achieved a temporal resolution of ~7 ms and a two-point threshold of 1.5 mm.
• Demonstrated superior classification accuracy for simple (99.0%) and complex (96.1%) patterns compared to humans (69.1% and 22.7%).
• The PRSA film enabled accurate pressure image acquisition for pattern analysis.

Conclusions:

• The developed artificial tactile finger significantly outperforms human tactile perception in pattern recognition.
• This technology offers a promising platform for advanced robotic tactile sensing.
• Potential applications include prosthetics, skin electronics, and robotic surgery.