Related Experiment Video
Updated: Aug 27, 2025
08:19
Asymmetric Walkway: A Novel Behavioral Assay for Studying Asymmetric Locomotion
Published on: January 15, 2016
8.9K
Accurate Real-time Phase Estimation for Normal and Asymmetric Gait
IEEE ... International Conference on Rehabilitation Robotics : [Proceedings]
|September 30, 2022
Summary
A new time-delay neural network accurately estimates gait phase in real-time for both normal and asymmetric walking. This gait phase estimator precisely detects heel strikes and assists movement with an exoskeleton.
Area of Science:
- Biomedical Engineering
- Robotics
- Machine Learning
Background:
- Accurate real-time gait phase estimation is crucial for developing effective human-robot interaction systems, particularly for lower-limb exoskeletons.
- Existing methods often struggle with the complexities of asymmetric and variable gait patterns.
Purpose of the Study:
- To develop and validate an accurate real-time gait phase estimator for both normal and asymmetric human locomotion.
- To assess the performance of the estimator in detecting key gait events like heel strikes.
- To evaluate the integration of the gait phase estimator with a spatial impedance controller for exoskeleton-assisted movement.
Main Methods:
- A time-delay neural network was trained and tested using gait data from six participants during treadmill walking.
- The estimator's accuracy was quantified using root mean square error (RMSE) and coefficient of determination (R²).
- A spatial impedance controller was implemented and tested using the estimated gait phase for exoskeleton assistance.
Main Results:
- The gait phase estimator achieved high accuracy, with RMSE < 3.48% for normal gait and < 4.31% for asymmetric gait.
- The coefficient of determination exceeded 99% for all subjects and gait types.
- Heel-strike detection showed precise results with RMSE < 2.56% (normal) and < 3.70% (asymmetric).
- The controller provided coordinated assistance during both normal and asymmetric gait.
- The estimator maintained accuracy across different walking conditions, including with exoskeleton use (passive and active modes).
Conclusions:
- The developed time-delay neural network provides a robust and accurate real-time gait phase estimation method.
- The estimator's precision in normal and asymmetric gait, along with heel-strike detection, supports its application in assistive robotics.
- The successful integration with a spatial impedance controller demonstrates the potential for seamless human-exoskeleton interaction.
