Enhanced decomposition of vertical ground reaction forces via an improved off-loading model and optimised artificial
Yifan Zhang1, Bintian Lin2, Changqing Miao3
1School of Engineering, Newcastle University, Newcastle upon Tyne NE1 7RU, UK; School of Civil Engineering, Southeast University, Nanjing 210096, China.
Gait & Posture
|July 23, 2026
Summary
This study presents an improved algorithm for accurately separating individual foot ground reaction forces (GRFs) during walking, even on uneven terrain. The new method significantly enhances precision in vertical GRF decomposition for practical biomechanical analysis.
Area of Science:
- Biomechanics
- Human Locomotion Analysis
- Gait Dynamics
Background:
- Accurate quantification of individual ground reaction forces (GRFs) is crucial for understanding human gait.
- A key challenge is decomposing GRFs during double support when both limbs contribute to the measured force.
- This difficulty is exacerbated by non-standard conditions like uneven surfaces or variable terrain.
Purpose of the Study:
- To enhance the accuracy of vertical GRF decomposition across diverse ground conditions.
- To develop a method for identifying unknown algorithmic constants for practical application.
- To refine the estimation of single-foot vertical GRF profiles.
Main Methods:
- Developed an improved two-term formulation based on the smooth-transition assumption for vertical GRF off-loading component estimation.
- Employed linear regression and artificial neural network (ANN) models, initialized with evolutionary computation, to predict model constants.
- Validated the method using experimental trials across various walking and platform-vibration conditions, and an open-access barefoot walking dataset.
Main Results:
- The proposed formulation achieved a 1.9% relative root mean square error for the off-loading component, a significant improvement over the original model's 12.2%.
- ANN and regression-based coefficient prediction yielded errors of 3.3% and 3.8%, respectively.
- Validation on a public dataset showed comparable low errors (2.0% for formulation, 3.0% for ANN).
Conclusions:
- The novel method effectively improves vertical GRF decomposition, even under challenging conditions like platform vibration.
- It offers a practical framework for accurately estimating single-foot vertical GRF from continuous force measurements.
- This advancement supports more precise biomechanical analyses of human locomotion.
