Using Fuzzy Neural Network Sliding Mode Control for Human-Exoskeleton interaction Forces Minimization

Abstract

Wearable robots incorporate human body with robotic machines into a single system, known as exoskeletons, through merging of human intelligence with robots' strength and durability. The development of exoskeleton for human performance augmentation necessitates the selection of actuation mode and a controlling method of human-robot interaction. The unidentified nonlinear dynamic properties of the systems or the uncertainties is a pressing matter in the model-based control of performance-augmenting exoskeleton systems. Uncertainties in the system may be caused by an inopportune approximation of the system dynamics that create considerable human-exoskeleton interaction forces while human motions. Therefore, robustness has to be necessarily incorporated by the controller of such exoskeleton systems in order to make it robust against the uncertainties. To control human-exoskeleton interactions, this study recommends a Fuzzy Neural Network Sliding Mode Control (FNNSMC) algorithm. Thereby, the proposed algorithm rules out the need for precisely estimated dynamic properties of the exoskeleton system, while the physical human-exoskeleton interaction forces reach further minimal values. Final performance verification of the recommended algorithm is carried out by simulation on 3-DOF exoskeleton swing led model. According to our simulation results, an appropriate control quality for the exoskeleton robot is provided by the proposed control algorithm. Moreover, a wearer carrying heavy load is supported by the exoskeleton, which concurrently tracks rapid movements of the wearer with no interference.