Predefined-time event-triggered adaptive neural practical tracking control for flexible-joint robot

Abstract

This paper researches the predefined-time event-triggered adaptive neural practical tracking control problem for flexible-joint robot system. An improved predefined-time command filter is utilized to get rid of the “explosion of complexity” problem, and the filter errors can be eliminated by an improved error compensation mechanism. Moreover, as an effective approximation tool, radial basis function neural networks are exploited to tackle the nonlinear terms existing in flexible-joint robot system. Finally, the communication burden of the system is reduced by using event-triggered technology, and an advanced adaptive predefined-time event-triggered controller is designed. The actual controller in the control scheme is updated only when it meets the conditions of the event-triggered mechanism. For flexible-joint robot system, the resulting control scheme makes the tracking error approach to a small neighborhood of zero, and all signals of the closed-loop system remain bounded within the predefined time. The effectiveness of the proposed controller is directly illustrated by the simulation results.