Ankle Moment Estimation Based on A Novel Distributed Plantar Pressure Sensing System.

Area of Science:

• Biomechanics
• Robotics
• Wearable Technology

Background:

• Ankle moment estimation is vital for gait analysis, rehabilitation, and exoskeleton robot control.
• Current methods using optical motion capture and force plates are lab-bound and impractical for real-world applications.
• A need exists for portable and accurate ankle moment estimation systems.

Purpose of the Study:

• To develop a novel distributed plantar pressure system for estimating ankle moment.
• To propose and evaluate machine learning models for ankle moment estimation using plantar pressure data.
• To validate the system's performance against gold-standard methods for overground walking.

Main Methods:

• Integration of eight pressure sensors into each insole to capture foot pressure distribution.
• Training of four neural network models, including a Genetic algorithm-Gated recurrent unit (GA-GRU), using plantar pressure data.
• Experimental validation with eight subjects during overground walking, using optical motion capture and force plates as references.

Main Results:

• The GA-GRU model demonstrated superior performance in generalized ankle moment estimation.
• The GA-GRU model achieved a normalized root mean square error (NRMSE) of 7.23% and a cross-correlation coefficient (ρ) of 0.85.
• The developed system and method proved effective for wearable robot control and human motion monitoring.

Conclusions:

• The novel distributed plantar pressure system offers a viable alternative for ankle moment estimation.
• The GA-GRU model provides a highly accurate and generalized approach for real-time ankle moment calculation.
• This technology has significant potential for applications in wearable robotics and human motion analysis.