Design and Validation of a Modified Delayed Output Feedback Controller for Hip Exoskeleton Assistance

Area of Science:

• Robotics
• Biomechanics
• Control Systems

Background:

• Conventional hip assistance controllers rely on gait phase estimation, limiting adaptability to activities like directional turns.
• Delayed Output Feedback Control (DOFC) offered phase-free torque but with limited profile complexity.

Purpose of the Study:

• To develop and evaluate a Modified Delayed Output Feedback Control (MDOFC) algorithm for enhanced energy efficiency in hip exoskeletons.
• To improve upon DOFC by enabling more adaptive and complex assistive torque profiles.

Main Methods:

• MDOFC was developed by incorporating harmonic components into the DOFC framework.
• Algorithm parameters were optimized using a reinforcement learning-based neural network controller within forward dynamic gait simulations.
• Performance was validated through dynamic stability analysis and human experiments.

Main Results:

• MDOFC demonstrated increased net metabolic cost reduction with only two parameters, comparable to DOFC.
• Dynamic stability was maintained up to 45 W, assessed by the short-term Lyapunov exponent.
• Human experiments showed MDOFC achieved similar metabolic benefits to DOFC while reducing mechanical power consumption by 8.8%.

Conclusions:

• MDOFC provides an effective control framework for hip exoskeletons, enhancing energy efficiency and adaptability.
• The algorithm's ability to generate adaptive torque profiles without explicit gait phase estimation is a key advancement.
• MDOFC represents a promising approach for developing next-generation wearable assistive devices.