Evaluating Electromyography and Sonomyography Sensor Fusion to Estimate Lower-Limb Kinematics Using Gaussian Process...

Area of Science:

• Robotics and Human-Machine Interaction
• Biomedical Engineering and Signal Processing
• Biomechanics and Assistive Technologies

Background:

• Advancements in robotic lower-limb assistive devices necessitate sophisticated control systems.
• Accurate intent recognition for these devices relies on robust sensing modalities.
• Surface electromyography (EMG) is a common but limited peripheral sensing method for muscle activity.

Purpose of the Study:

• To evaluate surface EMG and sonomyography (real-time ultrasound imaging of muscle) for predicting lower-limb joint kinematics.
• To assess the fusion of EMG and sonomyography features for enhanced control of robotic assistive devices.
• To investigate the continuous estimation of hip, knee, and ankle motion during diverse ambulation tasks.

Main Methods:

• Utilized Gaussian process regression, a Bayesian nonlinear model, for continuous estimation of joint angles and velocities.
• Extracted time-intensity features from anterior and posterior thigh sonomyography and time-domain features from lower-limb surface EMG.
• Trained and tested subject-dependent and task-invariant models for predicting lower-limb kinematics across level walking, stair, and ramp ambulation.

Main Results:

• Sensor fusion of anterior sonomyography with surface EMG significantly improved the estimation of hip, knee, and ankle motion for all tested ambulation tasks.
• Anterior sonomyography alone demonstrated significant error reduction at the hip and knee compared to surface EMG alone for most tasks.
• The findings indicate superior performance of sonomyography-based approaches over traditional EMG for lower-limb robotic control.

Conclusions:

• Combining sonomyography and EMG offers a more robust sensing solution for controlling complex robotic lower-limb assistive devices.
• Sonomyography, particularly from the anterior thigh, shows significant potential for improving volitional control strategies in assistive robotics.
• This research provides critical insights for the development and integration of advanced sensing modalities in next-generation robotic technologies.