Morse Code Recognition Based on a Flexible Tactile Sensor with Carbon Nanotube/Polyurethane Sponge Material by the...

Area of Science:

• Materials Science
• Human-Computer Interaction
• Machine Learning

Background:

• Morse code recognition is crucial for human-machine interaction.
• Developing precise tactile sensors for complex signal detection remains a challenge.

Purpose of the Study:

• To develop a flexible tactile sensor for accurate Morse code recognition.
• To evaluate the performance of a long short-term memory (LSTM) model in decoding Morse code signals.

Main Methods:

• Fabrication of a piezoresistive sensor using carbon nanotube (CNT) and polyurethane sponge (PUS) composite.
• Collection of 2160 voltage time-sequential signals for 36 Morse code types (letters A-Z, numbers 0-9).
• Application of data preprocessing (smoothing, normalization) and LSTM model for Morse code recognition, compared with GRU, SVM, MLP, and RF models.

Main Results:

• The CNT/PUS sensor precisely detected tactile features of Morse code.
• The LSTM model achieved high recognition accuracies: 99.17% for numbers, 95.37% for letters, and 93.98% for all 36 types.
• LSTM model outperformed GRU, SVM, MLP, and RF models in Morse code recognition accuracy.

Conclusions:

• The developed CNT/PUS sensor is effective for precise Morse code detection.
• The LSTM model demonstrates superior performance in recognizing Morse code signals from the tactile sensor data.