Gait Phase Recognition for Lower-Limb Exoskeleton with Only Joint Angular Sensors

Area of Science:

• Robotics
• Biomechanics
• Machine Learning

Background:

• Gait phase recognition is crucial for controlling lower-limb exoskeletons.
• Existing methods often rely on multiple sensor types, increasing complexity.
• Joint angular sensors are essential for closed-loop control and readily available.

Purpose of the Study:

• To develop a novel gait phase recognition method for lower-limb exoskeletons.
• To utilize only existing joint angular sensors, simplifying the system.
• To improve the accuracy and efficiency of gait phase identification.

Main Methods:

• Calculating and classifying gait deviation distances using Fisher's linear discriminant method.
• Dividing a gait cycle into eight distinct gait phases.
• Developing and training a multilayer perceptron model with phase-labeled gait data.

Main Results:

• The Fisher's linear discriminant method effectively classified gait deviation distances.
• The multilayer perceptron model achieved a 94.45% average correct rate of set (CRS).
• The model demonstrated an 87.22% average correct rate of phase (CRP) on the testing set, indicating accurate gait phase prediction.

Conclusions:

• The proposed method accurately recognizes gait phases using only joint angular sensors.
• This approach eliminates the need for additional sensors, simplifying exoskeleton systems.
• The findings contribute to more efficient and accessible lower-limb exoskeleton control.