This article proposes an adaptive robust controller based on neural networks (NNs) for industrial robot manipulators (IRM). In fact, robot manipulators belong to a nonlinear system and in the working process, they usually bear nonlinear fiction, payload variation, external disturbance, etc. To deal with these problems, an intelligent controller designed based on inheriting the advantages of the robust adaptive NNs and the SMC scheme to control the positions of industrial robot manipulators. In the proposed controller, the NNs are employed to approximate the unknown dynamic of the IRMs system. The adaptation laws of network parameters are established based on the Lyapunov stability theory as well as the guaranteed stability and robustness of the entire control system. Finally, simulations performed on a three-link robot industrial manipulator are provided in comparison with the Adaptive Fuzzy Control (AFC) and the Backstepping Control (BPC), thereby proving that the NNs controller can demonstrate superior tracking precision and higher robustness.