Online phase detection using wearable sensors for walking with a robotic prosthesis

Area of Science:

• Biomedical Engineering
• Rehabilitation Technology
• Human Movement Science

Background:

• Gait phase detection is crucial for effective control and feedback in robotic prostheses.
• Existing methods often require complex training data, limiting real-world application.
• Wearable sensor systems offer a promising avenue for unobtrusive gait analysis.

Purpose of the Study:

• To develop and evaluate a novel gait phase detection algorithm for robotic prosthesis users.
• To assess the algorithm's performance using a wearable wireless sensory system.
• To compare the algorithm's efficacy against established machine learning approaches.

Main Methods:

• Utilized a wearable wireless sensory system with sensorized shoe insoles and inertial measurement units.
• Implemented heuristic threshold rules to divide steady-state walking into four distinct gait phases.
• Conducted experiments with three lower-limb amputees using robotic prostheses and the sensor system.

Main Results:

• Achieved a high success rate (>90% average) in detecting all four gait phases across subjects.
• Demonstrated comparable performance to an off-line trained Hidden Markov Model (HMM) algorithm.
• Validated the algorithm's effectiveness without requiring prior learning dataset acquisition or model training.

Conclusions:

• The proposed heuristic-based gait phase detection algorithm is effective and accurate for robotic prosthesis users.
• This method offers a practical alternative to data-intensive machine learning models for real-time gait feedback.
• The system shows potential for enhancing user experience and rehabilitation outcomes with robotic prostheses.