Adaptive output feedback force tracking control for Lower Extremity Power-assisted Exoskeleton

Abstract

This paper presents an output feedback adaptive force tracking control scheme for Lower Extremity Power-assisted Exoskeleton (LEPEX). LEPEX is driven by electro-hydraulic actuators geometrically mounted for the active joints, which exhibits high nonlinear dynamics. In view of this, a robust observer resorting to Radius Basis Function Neural Network (RBF-NN) is proposed to approximate the nonlinearities so as to detect the values of the process variables.  An adaptive controller compatible with the RBF-NN estimator is adopted to cope with the nonlinearities and possible model uncertainties. The convergence of the output, i.e., load force variable, to a computable set is guaranteed in the context of Lyapunov direct method. Finally, simulation and experimental tests on the force control of an ankle joint of LEPEX are studied to witness the potentiality of the proposed approach.