Multi-Joint Motion Estimation of a Soft Gait Assistive Suit Using Pneumatic Reservoir Computing

In recent years, wearable power assistive devices to support human walking have been actively developed. A soft gait assistive suit actuated by pneumatic artificial muscles (PAMs) can detect gait phases from the pressure information of the PAMs. In the previous gait phase detection method of this suit, only a specific phase in the gait cycle has been detected. However, for more advanced assisting, it is desirable to be able to continuously detect the state of multiple parts of the leg during walking. In this study, we applied pneumatic reservoir computing, which is a computational method using the pressure responses of a flexible pneumatic pipeline system as a computational resource, to estimate the leg state of a soft gait assistive suit. As the state estimation of multiple parts of the leg, we attempted to continuously and simultaneously estimate the joint angles and joint angular velocities of the thigh and knee. The results showed that angles and angular velocities of multiple joints could be obtained continuously from a single pneumatic pipeline network with tubes that had adequate diameters and connected to multiple PAMs by using past pressure response data in the calculation.