Reducing the muscle activity of walking using a portable hip exoskeleton based on human-in-the-loop optimization

Area of Science:

• Robotics
• Biomechanics
• Human-Computer Interaction

Background:

• Human-in-the-loop optimization enhances wearable robotic devices for personalized assistance.
• Current methods require lengthy optimization periods, limiting practicality, especially for individuals with disabilities.

Purpose of the Study:

• To develop and validate a time-efficient muscle-activity-based human-in-the-loop optimization strategy for hip exoskeletons.
• To reduce the duration of biosignal collection during optimization iterations.

Main Methods:

• Implemented a muscle-activity-based optimization strategy using Bayesian and Covariance Matrix Adaptive Evolution Strategy (CMA-ES) algorithms.
• Utilized a portable hip exoskeleton to optimize rectus femoris muscle activity during walking.
• Conducted trials with four volunteers to assess the effectiveness of the optimization strategy.

Main Results:

• Reduced iteration time for biosignal collection from 120s to 25s.
• Achieved maximum muscle activity reductions of 33.56% (without exoskeleton) and 41.81% (with exoskeleton).
• Demonstrated that the muscle-activity-based strategy yielded superior assistance patterns compared to predefined ones for most participants.

Conclusions:

• The muscle-activity-based human-in-the-loop optimization strategy is effective and faster than metabolic cost-based methods.
• This approach offers a more practical and efficient way to personalize assistance from wearable robotic devices.
• The optimization results were consistent regardless of the order of Bayesian and CMA-ES algorithms used.