Human-in-the-Loop Trajectory Optimization Based on sEMG Biofeedback for Lower-Limb Exoskeleton.

Area of Science:

• Robotics
• Biomechanics
• Human-Machine Systems

Background:

• Lower-limb exoskeletons (LLEs) offer rehabilitation and walking assistance but often fail to account for individual differences.
• Personalization is key for intelligent human-machine systems (HMS) to optimize human performance.
• Human physiological response is a critical factor in trajectory optimization for LLEs.

Purpose of the Study:

• To propose a novel human-in-the-loop (HITL) motion planning method for LLEs.
• To enhance LLE adaptability and personalization by integrating biofeedback.
• To improve human performance and physiological response during exoskeleton-assisted locomotion.

Main Methods:

• Developed a hybrid dynamical model for the human-exoskeleton system.
• Employed offline trajectory optimization using direct collocation to create a gait library based on energy optimality.
• Integrated HITL trajectory selection using Thompson sampling with surface electromyography (sEMG) biofeedback.
• Implemented the selected trajectory using a hybrid zero dynamics control strategy.

Main Results:

• The proposed HITL method effectively optimizes gait trajectories for LLEs.
• Thompson sampling enabled personalized trajectory selection based on real-time biofeedback.
• Experimental validation demonstrated the effectiveness and superiority of the HITL approach compared to non-personalized methods.
• The system successfully adapted LLE behavior to individual user needs and physiological signals.

Conclusions:

• The HITL motion planning method significantly enhances LLE personalization and performance.
• Utilizing sEMG biofeedback is crucial for optimizing human-exoskeleton interaction.
• This approach offers a promising direction for intelligent LLEs in rehabilitation and assistive applications.