An iterative fuzzy controller for pneumatic muscle driven rehabilitation robot

Pneumatic muscle actuators (PMA) show great potential in wearable and compliant rehabilitation devices as they are flexible and lightweight. However, the varying and non-linear behavior of the actuators imposes modeling and control challenges, which are difficult to comprehend. This research proposes a new wearable ankle rehabilitation robot, first of its kind in the world driven by PMAs in a parallel form. The focus of this presented work is to develop an iterative controller to overcome the challenges for PMA driven devices. A fuzzy feedforward controller is proposed to accurately predict the behavior of PMA. A modified Genetic Algorithm (GA) is developed to identify the optimal set of parameters for the fuzzy controller. The iterative controller has been tested on the proposed PMA driven ankle rehabilitation robot, and is found capable of mapping the complex relationship in length, force and pressure of the PMA with high accuracy. Experimental results show excellent trajectory tracking performance of the controller when given various desired trajectories.

Research highlights
► We propose a new wearable robot for ankle rehabilitation, first of its kind driven by PMA.
► We examine the nonlinear behavior of PMA and develop a fuzzy fedforward controller to predict the behavior.
► The controller optimised using a modified genetic algorithm has shown improved trajectory tracking performance in our experiments.