Identification and Analysis of Human-Exoskeleton Coupling Parameters in Lower Extremities

Area of Science:

• Robotics
• Biomechanics
• Human-Machine Interaction

Background:

• Understanding human-exoskeleton coupling dynamics is crucial for developing effective and comfortable assistive devices.
• Existing models may not fully capture the complex interactions between humans and exoskeletons.

Purpose of the Study:

• To propose and validate linear and non-linear models for predicting human-exoskeleton coupling forces.
• To investigate correlations between coupling parameters, positions, and looseness.
• To develop a predictive model for human-exoskeleton coupling parameters without direct experimental measurement.

Main Methods:

• Development of linear and non-linear predictive models for coupling forces.
• Design of a novel experimental platform for parameter identification.
• Inclusion of ten adult males and ten adult females in experimental trials.
• Statistical analysis to determine correlations between coupling parameters and physical attributes.
• Application of backpropagation (BP) neural networks and Gaussian Process Regression (GPR) for parameter prediction.
• Sensitivity analysis of GPR performance to input parameters.

Main Results:

• The non-linear model demonstrated more accurate and robust prediction of coupling forces compared to the linear model.
• Significant correlations were identified between coupling parameters, positions, and looseness.
• GPR sensitivity analysis revealed the importance of individual input parameters for predicting coupling parameters.

Conclusions:

• The developed non-linear model offers superior prediction of human-exoskeleton coupling forces.
• The novel experimental platform and regression model facilitate obtaining human-exoskeleton coupling parameters.
• This research provides a foundation for optimizing control algorithms and designing more comfortable human-exoskeleton interfaces.