Virtual muscles and reflex control generates human-like ankle torques during gait perturbations

Area of Science:

• Biomechanics
• Robotics
• Neuroscience

Background:

• Achieving natural gait in powered prostheses and exoskeletons requires advanced control systems.
• Biologically-inspired actuation, incorporating muscles, spinal reflexes, and vestibular feedback, offers a promising avenue.

Purpose of the Study:

• To develop and evaluate a Virtual Muscle Reflex (VMR) system for controlling ankle torque.
• To compare the VMR system's performance against a conventional finite-state impedance controller in replicating human responses to perturbations.

Main Methods:

• Developed a real-time simulation of three Hill-Type muscles with feedback from ground reaction forces and virtual muscle stretch sensors.
• Optimized controller gains, muscle properties, and reflex/vestibular delays using Covariance Matrix Adaptation (CMA).
• Tuned the VMR system using human experimental data from anteroposterior mechanical perturbations at three walking speeds.

Main Results:

• The VMR system demonstrated lower root-mean-square error (RMSE) than the impedance controller in 70% of trials.
• The VMR controller achieved a higher coefficient of determination ($$) in 60% of trials compared to the impedance controller.
• The VMR system more accurately reproduced human torque responses to perturbations.

Conclusions:

• The Virtual Muscle Reflex (VMR) system shows superior performance in replicating human gait responses to perturbations compared to conventional impedance controllers.
• This biologically-inspired control strategy holds potential for improving the naturalness of movement in lower-limb assistive devices.
• Further development of VMR systems could lead to more intuitive and effective powered prostheses and exoskeletons.