Braille recognition by E-skin system based on binary memristive neural network.

Area of Science:

• Biomedical Engineering
• Materials Science
• Computer Science

Background:

• The Braille system is crucial for visually impaired individuals' communication.
• Certain groups, including young children, the elderly, and those with brain damage, face challenges learning Braille.
• A wearable, affordable Braille recognition system could significantly enhance accessibility and learning.

Purpose of the Study:

• To develop a low-cost, wearable electronic skin (E-skin) for Braille recognition.
• To create a system that assists visually impaired individuals in recognizing Braille and facilitates Braille learning.
• To demonstrate the feasibility of a cost-effective solution using advanced sensor technology and neural networks.

Main Methods:

• Fabrication of polydimethylsiloxane (PDMS)-based flexible pressure sensors to form an E-skin.
• Integration of the E-skin with a memristor-based neural network for Braille information processing.
• Implementation of a binary neural network algorithm with reduced computational complexity (two bias layers, three fully connected layers).

Main Results:

• The developed E-skin successfully mimics human touch for Braille data acquisition.
• The memristor-based neural network achieved a high Braille recognition accuracy of up to 91.25%.
• The simplified neural network design effectively reduced computational load and system costs.

Conclusions:

• A wearable and low-cost Braille recognition system is feasible using PDMS-based E-skin and memristor neural networks.
• This technology holds significant potential for assisting visually impaired individuals in Braille learning and communication.
• The study paves the way for more accessible assistive technologies for people with visual impairments.