Intelligent Internal Model Control of Robots for Upper-Limb Rehabilitation

This paper presents a neuro-fuzzy internal model Cartesian controller for robot manipulators. An inductive learning technique is applied to generate the required inverse dynamics and inverse kinematics modelling rules from input/output measurements. A fully differentiable fuzzy neural network is used to construct the adaptive sections of the controller for on-line parameters adaptation. A fuzzy-PID-like incremental controller is employed as feedback servo-controller. The internal model Cartesian controller is implemented using inverse kinematics and forward kinematics models of the robot. The proposed control system was tested using a dynamics model of a six-axis industrial robot to perform upper-limb rehabilitation. The obtained results demonstrate the validity of the proposed control scheme.