Near-Sensor Reservoir Computing for Gait Recognition via a Multi-Gate Electrolyte-Gated Transistor

Area of Science:

• Materials Science
• Electronics Engineering
• Biomedical Engineering

Background:

• Smart wearable electronics generate vast motion and physiological data.
• Conventional off-site processing causes latency and high energy consumption.
• Need for efficient on-device data analysis in wearables.

Purpose of the Study:

• To develop a near-sensor computing device for efficient multi-channel data processing.
• To utilize electrolyte-gated transistors (EGTs) for temporal feature extraction.
• To enable on-site analysis of motion and physiological signals in wearable systems.

Main Methods:

• A multi-gate electrolyte-gated transistor (EGT) based reservoir device was fabricated.
• EGTs were modulated by voltage to exhibit short-term dynamics.
• The EGT device was integrated with pressure sensors for gait analysis.

Main Results:

• The EGT device demonstrated nonlinear parallel integration of time-series signals.
• Temporal features like synchronization state and collective frequency were extracted.
• On-site gait analysis enabled identification of motion behaviors and Parkinson's disease.

Conclusions:

• The developed near-sensor reservoir computing system significantly improves efficiency.
• This approach offers a new pathway for rapid motion and physiological signal analysis.
• The technology paves the way for robust, flexible smart wearable electronics.