Gesture recognition in upper-limb prosthetics: a viability study using dynamic time warping and gyroscopes.

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Technology

Background:

• Current non-invasive interfaces for upper-limb prosthetics, like surface electromyography (sEMG), have limitations in fully utilizing advanced neuroprosthetic capabilities.
• The development of neuroprostheses aims to replicate human physiological performance, including dexterity and sensory feedback, necessitating improved interface solutions.

Purpose of the Study:

• To evaluate gesture recognition via gyroscope sensors as a viable non-invasive interface for upper-limb prosthetics.
• To assess the effectiveness of Dynamic Time Warping (DTW) for classifying prosthetic gestures.
• To identify optimal body sensor locations for gesture recognition in prosthetic control.

Main Methods:

• Utilized gyroscope sensors to capture upper-limb movements for gesture recognition.
• Applied Dynamic Time Warping (DTW) as a classification algorithm for gesture data.
• Investigated various sensor placements on the body to determine the most effective location.

Main Results:

• Achieved an optimal gesture classification rate of 97.53% (σ = 8.74).
• Identified that sensor placement proximal to the gesturing endpoint yields the highest classification accuracy.
• Demonstrated the potential of DTW for robust gesture recognition in this context.

Conclusions:

• Gesture recognition using gyroscope sensors presents a highly effective non-invasive interface for upper-limb prosthetics.
• The findings suggest that DTW is a suitable classification method for prosthetic gesture control.
• Optimal sensor placement is crucial for maximizing the performance of gesture-based prosthetic interfaces.