Identification of the best strategy to command variable stiffness using electromyographic signals

Area of Science:

• Robotics
• Human-Computer Interaction
• Biomedical Engineering

Background:

• Electromyography (EMG)-controlled robotic devices are increasingly common.
• Real-time stiffness control based on EMG signals can be fatiguing for operators.
• Novel control strategies are needed to mitigate fatigue in EMG-driven robotics.

Purpose of the Study:

• To propose and validate a novel stiffness control strategy for EMG-driven robotic devices.
• To investigate the impact of average muscle co-contraction on user fatigue and performance.
• To compare different averaging time windows (1s and 2s) for co-contraction estimation.

Main Methods:

• Nine subjects performed a wrist tracking task using a haptic device (Hi-5).
• Five sessions were conducted: Baseline, Proportional EMG control, Integral 1s, and Integral 2s co-contraction control.
• Task error, metabolic cost, perceived fatigue, and median EMG frequency were measured.

Main Results:

• All co-contraction estimation strategies positively impacted performance by reducing perturbation effects.
• The Integral 1s strategy resulted in lower metabolic cost compared to the Proportional strategy.
• The Integral 1s strategy led to lower perceived fatigue than both Proportional and Integral 2s strategies.

Conclusions:

• Controlling robotic device stiffness based on 1-second averaged muscle co-contraction is optimal.
• This strategy minimizes metabolic cost and perceived fatigue for the operator.
• The findings suggest a more sustainable and effective approach for EMG-driven robotic system design.