Reinforcement learning control for a robotic manipulator with unknown deadzone

Abstract

In this paper, an actor critic neural network control is developed for a robotic manipulator. Both system uncertainties and unknown deadzone are considered in the tracking control design. Stability of the closed-loop system is analyzed via the Lyapunov's direct method. The critic neural network is used to estimate the cost-to-go and the actor neural network is used to make the cost-to-go converge. Simulation studies are conducted to examine the effectiveness of the proposed actor critic neural network control.