Estimation of Knee Joint Angle from Surface EMG Using Multiple Kernels Relevance Vector Regression.

Area of Science:

• Robotics
• Biomedical Engineering
• Machine Learning

Background:

• Surface electromyography (sEMG) signals are crucial for motion intention recognition in wearable robots.
• Accurate knee joint angle estimation is vital for effective human-robot interaction and reducing model complexity.

Purpose of the Study:

• To propose and validate a novel Multiple Kernel Relevance Vector Regression (MKRVR) model for estimating knee joint angles using sEMG signals.
• To compare the performance of the MKRVR model against the Least Squares Support Vector Regression (LSSVR) model.

Main Methods:

• Offline learning utilizing Multiple Kernel Relevance Vector Regression (MKRVR).
• Performance evaluation using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and R2_score.
• Comparative analysis between MKRVR and Least Squares Support Vector Regression (LSSVR) models.

Main Results:

• The MKRVR model demonstrated superior performance in estimating knee joint angles compared to LSSVR.
• Achieved continuous global MAE of 3.27° ± 1.2°, RMSE of 4.81° ± 1.37°, and R² of 0.8946 ± 0.07.
• Validated the viability of MKRVR for accurate knee joint angle estimation from sEMG data.

Conclusions:

• The MKRVR method is a viable approach for estimating knee joint angles from sEMG signals.
• This technique can be applied to motion analysis and recognition of wearer's motion intentions in human-robot collaboration.
• Enhances the precision and reliability of control systems in wearable robotic applications.