Live magnetomyography and robotic-hand proxy control using a wearable triaxial optically pumped magnetometer

Area of Science:

• Biomagnetism
• Sensor Technology
• Human-Computer Interaction

Background:

• Optically pumped magnetometers (OPMs) offer ultra-sensitive, compact, contactless biomagnetic measurements.
• Measuring muscle activity for gesture recognition typically requires multiple sensors.

Purpose of the Study:

• To investigate the feasibility of using a single zero-field OPM to recognize and replicate multiple finger gestures.
• To assess the accuracy of a triaxial OPM system for classifying individual finger movements.

Main Methods:

• Simultaneous measurement of three-axis biomagnetic field components from forearm muscle activity during finger movements.
• Recording muscle activity for opening/closing each finger 60 times, triggered by auditory cues.
• Training a neural network for gesture classification and live replication of hand gestures onto a robotic hand.

Main Results:

• A single triaxial OPM successfully captured, recognized, and replicated finger gestures in a controlled setting.
• Three-finger gesture recognition and replication achieved ≥98% accuracy.
• Five-finger gesture recognition achieved ≥90% accuracy for four fingers, with the little finger at 45% accuracy.

Conclusions:

• A single zero-field OPM sensor with a triaxial measurement scheme is effective for contactless recognition and replication of multiple finger movements.
• This study provides a proof-of-concept for future wearable OPM-based magnetomyography systems.
• Further development is needed to improve accuracy for distinguishing subtle finger movements.