Empowering human-like walking with a bio-inspired gait controller for an under-actuated torque-driven human model

Area of Science:

• Biomechanics
• Robotics
• Computational modeling

Background:

• Accurate human gait simulation is vital for understanding locomotion, injury diagnosis, and developing rehabilitation technologies.
• Torque-driven human skeletal models in OpenSim offer anatomical accuracy but struggle with stabilizing unactuated degrees of freedom, like pelvis tilt, in forward dynamic simulations.

Purpose of the Study:

• To develop and evaluate a bio-inspired gait controller for torque-driven human skeletal models to achieve stable and energy-efficient walking.
• To address the challenge of stabilizing unactuated degrees of freedom in forward dynamic simulations.

Main Methods:

• A nonlinear model-based approach was used to calculate a balance-equivalent control torque.
• The hip-ankle strategy distributed this torque across lower-limb joints during the stance phase.
• A MATLAB-OpenSim interface facilitated forward dynamic simulations for optimizing controller parameters and evaluating robustness.

Main Results:

• The torque-driven model achieved natural gait with joint torques closely matching experimental data.
• The bio-inspired controller demonstrated robustness against external forces, showing efficient balance recovery.
• Simulations confirmed stable walking and minimized energy expenditure.

Conclusions:

• The proposed bio-inspired gait controller effectively enables stable walking in torque-driven human skeletal models.
• This approach enhances the realism of gait simulations and has potential applications in rehabilitation and assistive technology development.
• The controller's robustness to external disturbances highlights its practical applicability for realistic locomotion analysis.