Real-Time Gait Phase and Task Estimation for Controlling a Powered Ankle Exoskeleton on Extremely Uneven Terrain

Area of Science:

• Biomechanics
• Robotics
• Human-Computer Interaction

Background:

• Lower-limb exoskeletons show promise in lab settings but struggle with real-world gait adaptation.
• Existing controllers lack real-time synchronization with dynamic human locomotion and varying environmental conditions.

Purpose of the Study:

• To develop and validate a novel controller for ankle exoskeletons capable of real-time gait phase and task variable estimation.
• To enable adaptive torque assistance that synchronizes with human gait across diverse activities and terrains.

Main Methods:

• A data-driven kinematic model was employed to continuously estimate gait phase, phase rate, stride length, and ground incline.
• Torque assistance was adapted in real-time based on a multi-activity database of human locomotion data.
• Live experiments were conducted with able-bodied participants on treadmills and uneven terrain.

Main Results:

• The controller achieved phase estimates comparable to state-of-the-art methods.
• Task variable estimation accuracy was similar to recent machine learning approaches.
• Successful adaptation of assistance was demonstrated during controlled and real-world locomotion tasks, with low phase Root Mean Square Error (RMSE).

Conclusions:

• The developed controller effectively enables ankle exoskeletons to adapt assistance in real-time to human gait dynamics.
• This advancement facilitates more natural and effective exoskeleton use in complex, real-world environments.
• The controller shows significant potential for improving mobility assistance devices.