Multi-Modal Muscle Activation Modeling Using Koopman Operator Linearization for an Ankle Exoskeleton

Area of Science:

• Biomechanics
• Robotics
• Rehabilitation Engineering

Background:

• Gait impairments significantly reduce mobility and quality of life.
• Intelligent rehabilitation devices require dynamic models for effective human-machine interaction and assistance.
• Accurate estimation of volitional torque is crucial for personalized gait support.

Purpose of the Study:

• To develop a data-driven model for estimating volitional ankle torque during the gait cycle.
• To integrate Koopman operator linear estimations with fused surface electromyography (sEMG) and ultrasound (US) imaging.
• To improve the accuracy of state and volitional torque estimation for enhanced gait rehabilitation.

Main Methods:

• Utilized Koopman operator linear estimations to model nonlinear, phase-dependent ankle motion dynamics.
• Fused surface electromyography (sEMG) and ultrasound (US) imaging data to enhance muscle activation predictions.
• Developed a data-driven framework for real-time volitional torque estimation throughout the gait cycle.

Main Results:

• The integrated model demonstrated improved accuracy in estimating volitional ankle torque.
• Fusion of sEMG and US signals provided combined electrical and morphological insights into muscle activation.
• The framework successfully estimated nonlinear, phase-dependent dynamics of ankle motion.

Conclusions:

• The proposed data-driven model shows significant potential for improving gait rehabilitation strategies.
• Accurate volitional torque estimation is achievable through the fusion of sEMG, US imaging, and Koopman operator methods.
• This framework lays the foundation for individualized exoskeleton assistance and more effective gait rehabilitation.