Closing the Loop on Exoskeleton Motor Controllers: Benefits of Regression-Based Open-Loop Control.

Area of Science:

• Robotics
• Biomechanics
• Control Systems Engineering

Background:

• Lower-limb exoskeletons enhance walking performance and mobility.
• Sensor-less, open-loop motor control offers reduced complexity and cost for consumer applications.
• Developing effective open-loop controllers for exoskeletons is challenging due to variable transmission systems.

Purpose of the Study:

• To develop and validate an open-loop motor control framework for lower-limb exoskeletons.
• To achieve performance comparable to or exceeding closed-loop torque control.
• To investigate sensor-less control for improved consumer exoskeleton applications.

Main Methods:

• Generalized linear regression was used to model motor current during exoskeleton-assisted walking.
• Two open-loop controllers were developed: a 'complex' model (desired torque, estimated ankle angular velocity) and a 'simple' model (desired torque only).
• Five participants walked on a treadmill and over-ground with closed-loop and open-loop controllers providing ankle exoskeleton assistance.

Main Results:

• Open-loop controllers demonstrated similar root-mean-squared torque tracking error compared to closed-loop control.
• Both open-loop controllers showed improved average torque production, lower power consumption, and reduced operating noise versus closed-loop control.
• New control models validated on a different exoskeleton configuration replicated these improvements during over-ground walking.

Conclusions:

• The developed open-loop motor control framework effectively matches closed-loop control performance in exoskeletons.
• Sensor-less, open-loop control presents a viable and advantageous alternative for exoskeleton motor control, particularly for consumer applications.
• This framework enables the creation of responsive and efficient exoskeleton controllers with reduced system complexity.