Efficient correction of armband rotation for myoelectric-based gesture control interface

Area of Science:

• Biomedical Engineering
• Rehabilitation Engineering

Background:

• Commercial myoelectric armbands enhance portability of myoelectric controlled interfaces (MCIs).
• Current MCIs lack robustness against armband displacement, particularly rotation, causing performance degradation.
• Retraining MCIs after displacement is impractical due to extensive data collection requirements.

Purpose of the Study:

• To validate the online effectiveness of the position verification (PV) framework for MCIs.
• To introduce a novel algorithm for identifying rotation direction to enhance PV framework efficiency.
• To demonstrate the real-time capability of the PV framework with a commercial armband.

Main Methods:

• Development of a novel algorithm to identify armband rotation direction.
• Implementation and real-time testing of the position verification (PV) framework.
• Evaluation of PV framework's effectiveness in correcting electrode shifts on a commercial armband.

Main Results:

• The PV framework corrected a 1.5-cm armband rotation with an average of 3.1 ± 1.5 adjustments (15.5 ± 7.5 s).
• No significant difference in real-time control performance was observed before displacement and after PV correction.
• The study demonstrated the real-time capability of the PV framework for myoelectric control.

Conclusions:

• The PV framework significantly reduces correction time compared to retraining.
• This study is the first to maintain pattern recognition-based myoelectric control performance despite electrode shifts without retraining.
• The findings suggest the feasibility of the PV framework for practical applications of myoelectric armbands and MCIs.