The Fifth Main Dynamic Factor: Skin Temperature and Its Effects on EMG-Based Gesture Recognition

Area of Science:

• Biomedical Engineering
• Human-Computer Interaction
• Signal Processing

Background:

• Surface electromyography (sEMG) is vital for gesture recognition in prosthetics and human-machine interfaces.
• sEMG performance is affected by factors like limb position, electrode shift, and signal non-stationarities.

Purpose of the Study:

• To investigate skin temperature as a novel dynamic factor influencing sEMG-based gesture recognition.
• To evaluate the impact of varying skin temperatures on sEMG signal characteristics and gesture classification accuracy.

Main Methods:

• Collected synchronized sEMG, photoplethysmography (PPG), and inertial measurement unit (IMU) data from 12 participants across cold, baseline, and hot conditions.
• Assessed sEMG signal features (mean absolute value, median frequency, fuzzy entropy) and gesture recognition model performance under different temperatures.
• Utilized the BioPoint wearable device for data acquisition.

Main Results:

• Cold temperatures altered sEMG signal characteristics (increased mean absolute value, decreased median frequency).
• Gesture recognition models trained on baseline data performed poorly in non-baseline temperature conditions.
• Training with temperature-diverse data improved robustness but slightly decreased baseline accuracy.
• Multisensor approaches (EMG, PPG, IMU) enhanced baseline accuracy but increased sensitivity to temperature changes.

Conclusions:

• Skin temperature is a critical, previously overlooked factor in sEMG-based gesture recognition.
• Developing temperature-aware strategies is essential for robust and adaptive human-machine interfaces and prosthetic control.
• A new hand gesture dataset under varying temperatures is introduced to facilitate future research.