Neural Adaptive Integral-Sliding-Mode Controller with a SSVEP-based BCI for Exoskeletons

Abstract

This paper introduces a robust neural adaptive integral sliding mode controller with a SSVEP-based BCI for exoskeletons. A BCI is used to establish the desired trajectories by analyzing EEG signals. The neural networks are used to approximate nonlinear exoskeleton's dynamic. A sliding mode controller is added to guarantee the global asymptotic stability of the tracking trajectory and the neural network approximations. The controller's design is based on the hypothesis that only classical properties like boundedness of some parameters are known and all other functions are unknown. The closed-loop stability of the system is demonstrated using Lyapunov method. The effectiveness of the proposed approach is tested by an experiment application to rehabilitation context using an upper limb exoskeleton of 2-DOF.