Adaptive neural network control of flexible joint manipulators in constrained motion

Abstract

This paper addresses the control problem of flexible joint manipulators in constrained motion. The singular perturbation method is applied to reduce the original system into slow and fast subsystems. By adopting the composite control strategy, an adaptive neural network controller is first developed to control the slow subsystem. Then a fast feedback control is designed to stabilise the fast subsystem along its equilibrium trajectory. The system stability is proven using Lyapunov theory. It is shown that the motion tracking error converges to zero asymptotically, whereas the force tracking error remains bounded and can be made arbitrarily small. Numerical simulations are provided to show the effectiveness of the approach.