Multijoint Continuous Motion Estimation for Human Lower Limb Based on Surface Electromyography

Sensors, Journal Year: 2025, Volume and Issue: 25(3), P. 719 - 719

Published: Jan. 24, 2025

The estimation of multijoint angles is great significance in the fields lower limb rehabilitation, motion control, and exoskeleton robotics. Accurate joint angle helps assess function, assist rehabilitation training, optimize robotic control strategies. However, estimating different movement patterns, such as walking, obstacle crossing, squatting, knee flexion–extension, using surface electromyography (sEMG) signals remains a challenge. In this study, model proposed for continuous (CB-TCN: temporal convolutional network + block attention module network). integrates networks (TCNs) with modules (CBAMs) to enhance feature extraction improve prediction accuracy. effectively captures features movements, while enhancing key through mechanism CBAM. To model’s generalization ability, study adopts sliding window data augmentation method expand training samples adaptability patterns. Through experimental validation on 8 subjects across four typical results show that CB-TCN outperforms traditional models terms accuracy robustness. Specifically, achieved R2 values up 0.9718, RMSE low 1.2648°, NRMSE 0.05234 during walking. These findings indicate combining TCN CBAM has significant advantages predicting angles. approach shows promise analysis.

Language: Английский

Lower Limbs 3D Joint Kinematics Estimation From Force Plates Data and Machine Learning

Published: Nov. 22, 2024

This study investigated the possibility of using machine learning to estimate 3D lower-limb joint kinematics during a rehabilitation squat exercise from force plate data, that can be collected very simply outside laboratory and does not pose privacy issues. The proposed approach is based on bidirectional-Long-Short-Term-Memory (Bi-LSTM) associated Multi-Layer-Perceptron (MLP) model. use MLP allows fast training evaluation time. model was trained validated nineteen healthy young volunteers stereophotogrammetric motion capture system collect ground truth data. Volunteers performed squats in normal conditions an ankle brace simulate pathological motion. Also additional loads were added onto lower limbs segments influence atypical mass distribution. root mean square differences between estimated angles those reconstructed with than 6deg correlation coefficients higher 0.9 average. Furthermore, inference time as low $12 \mu \mathrm{~s}$ paving way future reliable real-time measurement tools.

Language: Английский