Extended-state-observer-based adaptive control of flexible-joint space manipulators with system uncertainties

In this paper, an extended-state-observer-based adaptive controller is proposed for flexible-joint space manipulators (FJSM) to accurately track trajectories while stabilizing bases in the presence of dynamic uncertainties and joint stiffness uncertainties. The dynamic model of a FJSM is established, and its state-spaced representation is obtained by introducing an error vector and a sliding mode surface vector as state variables. Besides, an extended state observer (ESO) is designed to guarantee the precise estimation of the manipulator's velocity states as well as the joint stiffness uncertainties. Based on the ESO and the state-spaced representation, an adaptive controller is generated by implementing backstepping method, where the dynamic uncertainties are compensated by a Radial Basis Function neural network (RBFNN) and the joint stiffness uncertainties are eliminated by the estimation of the ESO. The stabilities of the ESO-based adaptive controller are validated by Lyapunov theory. Several numerical simulations were conducted, and the simulation results verifies the effectiveness of the proposed controller. (c) 2022 Published by Elsevier B.V. on behalf of COSPAR.

Automated summary

This paper proposes an extended-state-observer-based adaptive controller for flexible-joint space manipulators (FJSM) to accurately track trajectories while stabilizing bases in the presence of dynamic uncertainties and joint stiffness uncertainties. The dynamic model of a FJSM is established, and an extended state observer (ESO) is designed to estimate the manipulator's velocity states and joint stiffness uncertainties. An adaptive controller is generated based on the ESO and the state-spaced representation, which compensates for dynamic uncertainties using a Radial Basis Function neural network (RBFNN) and eliminates joint stiffness uncertainties through the ESO estimation. The stabilities of the ESO-based adaptive controller are validated using Lyapunov theory, and numerical simulations confirm the effectiveness of the proposed controller.