Fixed-Time Prescribed Performance Control for Robotic Manipulators via Reinforcement Learning

Abstract

In this article, a reinforcement learning (RL)-based fixed-time trajectory tracking control scheme is proposed for robotic manipulators, where the unknown disturbances and model uncertainties are considered. A nonsingular fast terminal sliding surface is designed to guarantee the fixed-time convergence of the tracking error, with an estimation of the upper limit for the convergence time provided. A performance index function is formulated, incorporating both the tracking error and input cost. By employing an adaptive identifier and a critic, we develop an RL-based optimal tracking control scheme, where the convergence time is independent of initial conditions. The stability of the system and the fixed-time convergence of the closed-loop system are demonstrated via Lyapunov theory. Physical simulations and experiments are carried out to verify the feasibility of the proposed control strategy.