2026 Volume 57 Issue 1 Pages 9-12
This paper proposes a pneumatic assist suit that achieves posture estimation and gait assistance in a single system through gait-phase classification based on the estimated hip angular velocity. The proposed method switches the gait phase and control law based on the zero-crossing timing of the estimated hip angular velocity. Because the zero-crossings of hip angular velocity correspond to extrema of the hip angle, the proposed method eliminates the need for participant-specific tuning of angle thresholds for gait-phase classification. During phases in which a pneumatic artificial muscle (PAM) is not used for assistance, it is depressurized and closed at low pressure, and the pressure time series generated by PAM length changes during walking is used to estimate the hip angle and angular velocity through random feature mapping and ridge regression. Treadmill walking experiments with two healthy male participants confirmed that posture estimation based on system pressure time series, gait-phase classification based on the estimated posture, and PAM-based gait assistance can be executed in real time within the same pneumatic system. The results also showed that posture estimation performance and the stability of zero-crossing timing detection were mostly maintained even under conditions involving role switching of the PAMs.
