Preliminary Study on Walking Gait Classification Utilizing Ankle Position and Ankle Velocity

This research conducts a preliminary study on ankle position and ankle velocity-based walking gait classification to be implemented in a Passive Controllable Ankle Foot Orthosis (PICAFO). Ankle position and ankle velocity data was taken from a single subject walks in 3 different walking speed (1, 3, and 5 km/h). Support Vector Machine (SVM) is used as the classification method. The SVM was trained using 75% of the data and was tested using 25% of the data. Three types of Gaussian Kernel were compared for better accuracy, such as fine Gaussian (γ = 0.35), medium Gaussian (γ = 1.4), and Coarse Gaussian (γ = 5.7). The accuracy is observed in several cases based on walking gait data in each walking speed or all the walking gait data combined. The results show that the SVM method of walking gait classification based on ankle position and ankle velocity is promising. The fine gaussian SVM (γ = 0.35) produces the highest walking gait classification accuracy of 88.4%. In the future, the study should confirm the possibility of the proposed classification method for the non-single subject.