Adaptive Control of Upper Limb Exoskeleton Robot Based on a New Modified Function Approximation Technique (FAT)

To improve the performance of passive-assistive rehabilitation motion with the help of an exoskeleton robot, an adaptive tracking controller based on Function Approximation Technique combined with disturbance observer, taking into consideration the actuators dynamics is presented. The control scheme aims to approximate the dynamic model of the robot and provides an accurate compensation of non-smooth nonlinear constraints that are excited by backlash, hysteresis, deadzone, and saturation of the robot's actuators. Unlike a conventional Function Approximation Technique approach, the required use of basis functions in estimation’s law of the dynamic model is eliminated and the acceleration feedback is also eliminated. The output of the disturbance observer is linked directly to the current loop, which permits to the control system to be strong and active to eliminate the nonlinear constraints effects. Thanks to the proposed strategy, the designed control approach requires no knowledge of dynamic parameters of the exoskeleton robot, including the dynamics of its actuators to achieve the desired performance. Simulation results and a comparative study are presented to validate the proposed approach.