Using the Intact Human Hand to Benchmark Real-Time Myoelectric Control Performance for Robotic Interfaces

Area of Science:

• Biomedical Engineering
• Robotics
• Neuroscience

Background:

• Electromyogram (EMG)-based myoelectric control is crucial for advanced robotic interfaces like prostheses.
• Establishing a benchmark for EMG controller performance is essential for clinical translation.

Purpose of the Study:

• To evaluate the intact human hand as a gold standard for assessing EMG-based myoelectric control systems.
• To compare the performance of a musculoskeletal model-based EMG controller against natural human hand control.

Main Methods:

• A real-time virtual posture-matching task was performed by participants using both their intact hand (goniometer trials) and an EMG controller (model trials).
• Within-subjects comparison of normalized path length, task duration, joint angle accuracy, and variability.

Main Results:

• Goniometer trials demonstrated significantly better performance with lower normalized path length (2.0±1.6 vs 4.1±4.3) and shorter task duration (3.3±3.4 sec vs 12.3±10.7 sec) compared to model trials (p<0.0001).
• EMG controller trials exhibited greater joint angle variability and a constant offset between actual and virtual joint postures.

Conclusions:

• The intact human hand provides a superior benchmark for myoelectric control compared to current EMG-based systems, including musculoskeletal model controllers.
• Quantifying performance differences informs advancements in EMG control algorithms for more intuitive and effective robotic interfaces.
• The intact hand serves as an objective standard for selecting EMG control strategies for clinical applications.