Chia, En YuEn YuChiaChang, Yuan KaiYuan KaiChangChang, Yao ChengYao ChengChangChen, Yi LianYi LianChenChien, Tzu ChiehTzu ChiehChienChiang, Ming LiMing LiChiangLI-CHEN FUJIN-SHIN LAILU LU2024-01-122024-01-122024-02-0109574158https://scholars.lib.ntu.edu.tw/handle/123456789/638459Assist-As-Needed (AAN) rehabilitation in related works typically limits subjects to perform rehabilitation tasks with particular time instants. This paper proposes a velocity field based on an active-assistive control system to solve this issue. The active intention of motion of the subjects are extracted via a Kalman filter constructed on the basis of an interactive torque observer. Then, given a task motion pattern devised to provide subjects the time-decoupled assistance, our system can first automatically produce a position-coupled velocity field. Afterwards, a method is proposed to integrate the active and assistive motion based on the subjects’ performance and involvement so that eventually they will actively perform the task in a more accurate manner. The stability of the proposed system is verified by Lyapunov analysis. The experiment results show that the execution time and the subjects’ exertion are reduced when performing given tasks compared with the related work. To sum up, the hereby developed system not only can maintain the subjects’ active intention but also can assist them to run the task accurately.Active-assistive rehabilitation | Assist-as-needed rehabilitation | Contact force estimation | Rehabilitation roboticsjournal article10.1016/j.mechatronics.2023.1031152-s2.0-85179845563https://api.elsevier.com/content/abstract/scopus_id/85179845563