IJCAS

Robot-assisted rehabilitation systems have shown promising advantages over traditional therapist-based methods. The type of the controller has an important role in the efficiency of such systems. In this regard, this paper presents a new assist-as-needed (AAN) controller for 4-cable planar robots. The main purpose is to design a bounded-input AAN controller with an adjustable assistance level and a guaranteed closed-loop stability. The proposed controller involves the advantages of both the model-based and non-model-based AAN controllers, and in this way can increase the efficiency of rehabilitation. The controller aims to follow a desired trajectory by allowing an adjustable tracking error, which enables the human subject to freely move the target limb inside this error area. This feature of the controller gives an important advantage over the existing model-based controllers. The controller also compensates for the dynamic modeling uncertainties of the system through an adaptive neural network. The adaptive term includes a forgetting factor to adjust the assistance level of neural network term. The stability of the closed-loop system is analysed, and the uniformly ultimately bounded stability is proven. The effectiveness of the proposed control scheme is validated through simulations conducted for gait rehabilitation."

Keywords: Adaptive control, assist-as-needed control, neural network, parallel robot, rehabilitation.